Due to the life cycle of its vector, Lyme disease has known seasonal variation. However, investigations focused on children have been limited. Our objective was to evaluate the seasonality of pediatric Lyme disease in three endemic regions in the United States. We enrolled children presenting to one of eight Pedi Lyme Net participating emergency departments. Cases were classified based on presenting symptoms: early (single erythema migrans (EM) lesion), early-disseminated (multiple EM lesions, headache, cranial neuropathy, or carditis), or late (arthritis). We defined a case of Lyme disease by the presence of an EM lesion or a positive two-tier Lyme disease serology. To measure seasonal variability, we estimated Fourier regression models to capture cyclical patterns in Lyme disease incidence. While most children with early or early-disseminated Lyme disease presented during the summer months, children with Lyme arthritis presented throughout the year. Clinicians should consider Lyme disease when evaluating children with acute arthritis throughout the year.
T-optimal discriminating designs for Fourier regression models
