ABSTRACT
BACKGROUND: The utility of peripheral blood cultures in pediatric oncology patients presenting with fever is controversial. A recent systematic review showed that about one in 40 bloodstream infections (BSIs) would be missed if only central venous line (CVL) cultures are obtained. OBJECTIVE: To derive a clinical decision rule for obtaining peripheral blood cultures in pediatric oncology patients presenting to a pediatric emergency department (PED) with fever and a CVL. DESIGN/METHOD: A retrospective chart review was performed on pediatric oncology patients referred to the PED for fever while on therapy. Logistic regression with a random intercept was used to determine independent predictors of BSI and generate a prediction model for obtaining peripheral blood cultures. The decision rule was generated from the best performance as measured by a receiver operator curve. Bootstrapping analysis was performed for internal validation. RESULTS: Predictors that were significant and independently associated with positive peripheral blood cultures included vasopressor support (odds ratio [OR] 16.5, 95% confidence interval [CI]: 2.80-97.71), acute myeloid leukemia (AML) diagnosis (OR 6.9, 95% CI: 1.81-25.98), hypotension (OR 4.0, 95% CI: 1.05-15.17), mucositis (OR 8.2, 95% CI: 2.48-27.01), and maximum temperature in PED ≥39°C (OR 6.6, 95% CI: 2.36-18.20). The area under the curve (AUC) for this model was 0.90 (95% CI: 0.82-0.97) in the derivation cohort and 0.90 (95% CI: 0.81-0.98) after the internal validation. CONCLUSIONS: We derived a clinical prediction model for deciding when to obtain peripheral blood cultures in febrile oncology patients with CVLs on active therapy. Future studies should focus on prospective and external validation of this diagnostic prediction tool.
