Investigators from the University of Alabama at Birmingham conducted a retrospective observational study to identify predictors of bacteremia and readmission in febrile children with sickle cell anemia (SCA). Patients with SCA who were <18 years old and presented to the study ED between 1999–2016 with a temperature of ≥101°F were eligible. The standard of care at the study ED during the study period was to obtain a complete blood count and blood culture prior to administering a dose of antibiotics, with hospital admission being at the discretion of the treating physician, except for SCA patients with fever and an abnormal chest x-ray (who were always admitted per study ED policy). Patients with an abnormal chest x-ray were therefore excluded from the study. Investigators abstracted demographics, laboratory results, and presenting clinical signs and symptoms of eligible patients from their medical chart.

The primary outcome was bacteremia, defined as a positive blood culture. The primary predictor variables included ill appearance (defined by presence of “uncomfortable,” “chills,” “lethargic,” or “seizure” in ED physician documentation), high temperature (≥39.5°C), abnormal heart rate, abnormal blood pressure, and white blood cell (WBC) count of >30 or <5 x 10⁹/L on presentation. A secondary outcome was readmission within 14 days from initial presentation among those initially discharged from the ED or from the hospital within 72 hours of their admission. Investigators used logistic regression to identify predictor variables associated with each outcome as well as receiver operator curves (ROC) using different sets of predictor variables to assess sensitivity and specificity in detecting bacteremia.

There were 155 patients included in analysis who had a total of 653 febrile events. Of these events, 78% resulted in hospitalization and 2.6% (N=17) were associated with bacteremia. Readmission occurred in 6% (N=26) of patients with febrile events discharged from the ED or discharged from the hospital within 72 hours of admission.

The only variable associated with bacteremia was ill appearance (odds ratio, 8.5; 95% CI, 2.85, 25.33). The best-performing ROC models for bacteremia included: (1) ill appearance, WBC count >30 x 10⁹/L, or hypotension (sensitivity 71%, specificity 84%), or (2) ill appearance, WBC count >20 x 10⁹/L, or hypotension (sensitivity 88%, specificity 64%). There were no significant differences in predictor variables between febrile events that required readmission and those that did not.

The investigators conclude that bacteremia is rare in febrile children with SCA and that admission criteria can be based on ill appearance, WBC count, and hypotension.