Objective To identify characteristics differentiating the node-first presentation of Kawasaki disease (NFKD) from bacterial cervical lymphadenitis (BCL) and typical Kawasaki disease (KD). Study design From our prospectively collected database, we compared clinical, laboratory, and imaging characteristics of NFKD and BCL cohorts and performed multivariable logistic regression to identify variables that distinguish NFKD from BCL. We then compared outcomes of NFKD and patients with typical KD treated during the same period. Results Over 7 years, 57 patients were hospitalized for NFKD, 78 for BCL, and 287 for typical KD. Patients with NFKD were older and had more medical encounters and longer duration of illness prior to correct diagnosis than patients with BCL. Of patients with NFKD, 33% had an admission diagnosis of bacterial adenitis or abscess. Compared with patients with BCL, patients with NFKD had lower leukocyte (WBC), hemoglobin, and platelet counts and higher absolute band counts (ABC), C-reactive protein (CRP), alanine transaminase and γ-glutamyl transpeptidase levels, and erythrocyte sedimentation rates. In the multivariable analysis, smaller nodes, lower WBC, and higher ABC and CRP were independently associated with NFKD. Patients with NFKD had multiple enlarged solid nodes and comparable rates of retropharyngeal edema. Compared with patients with typical KD, patients with NFKD were older, had more severe inflammation, and had similar rates of coronary artery abnormalities and resistance to intravenous immune globulin. Conclusions High ABC and CRP values and multiple enlarged solid nodes in febrile patients with cervical adenopathy should prompt consideration of NFKD to prevent delayed diagnosis of KD. Retropharyngeal edema on radiography should not dissuade from the diagnosis of NFKD.

Overcrowding and prolonged patient length-of-stay (LOS) in emergency departments (EDs) are growing problems. We evaluated the impact of implementing a rapid whole blood quantitative D-dimer test (Biosite Triage, Biosite Diagnostics, San Diego, CA) in our ED satellite laboratory on 252 patients before vs 211 patients after implementation. All patients also underwent testing with the existing central laboratory method (VIDAS D-dimer, bioMérieux, Durham, NC). D-dimer turnaround time (from blood draw to result) decreased approximately 79% (approximately 2 hours vs 25 minutes). The mean ED LOS declined from 8.46 to 7.14 hours (P = .016). Hospital admissions decreased 13.8%, ED discharges increased 7.3%, and the number of patients admitted for observation increased 6.4% (P = .005). No difference in the utilization of radiologic studies was observed (P = .86). At 3 months' follow-up, none of the after-implementation patients with negative D-dimer results were admitted for subsequent venous thromboembolic disease. The rapid D-dimer test was associated with a shorter ED LOS and fewer hospital admissions.

No abstract