Use of Risk Prediction Model for Bloodstream Infection in Febrile Pediatric Patients With Cancer Without Severe Neutropenia

Get Permission

As reported in the Journal of Clinical Oncology, Esbenshade et al prospectively evaluated the Esbenshade Vanderbilt (EsVan) model for risk prediction of bloodstream infections in febrile pediatric patients with cancer without severe neutropenia, showing that the model accurately predicted bloodstream infections and permitted safe reduction of antibiotic use in this population. 

Study Details

Data were prospectively collected from febrile pediatric patients with a central venous catheter and absolute neutrophil count ≥ 500/µL from April 2015 to August 2019 at a single site (Monroe Carrel Junior Children’s Hospital at Vanderbilt University). The initial EsVan model, which contains 12 clinical variables, was used from 2015 to 2017; the subsequent EsVan2 model, which eliminated two of the variables, was used from October 2017 to 2019.

Patients with low bloodstream infection risk (< 10%) using the EsVan models were to be discharged home without antibiotics, those with intermediate risk (10%–39.9%) were to be given an antibiotic before discharge, and those with high risk (≥ 40%) were to be admitted on broad-spectrum antibiotics. Seven-day outcomes were collected, and the accuracy of the models was evaluated using C-statistics.

Key Findings

The EsVan models were used in 755 (80.6%) of the total of 937 febrile, nonsevere neutropenia episodes in 331 patients. Among the 937 episodes, 88.9% were low risk, 8.6% were intermediate risk, and 2.3% were high risk based on model predictions.

In the overall population, bloodstream infection occurred in 4.2% of episodes, including 1.9% of low-risk, 13.6% of intermediate-risk, and 54.5% of high-risk episodes. In the low-risk group, empirical intravenous (IV) antibiotics were administered in 21.1% of cases, with oral antibiotics given in an additional 8.8%. By 7 days postpresentation, 72.3% of episodes never required IV antibiotics and 62.4% required no IV or oral antibiotics. No deaths or clinical decompensation were attributable to antibiotic delay.

In the entire cohort, the C-statistic for bloodstream infection detection was 0.802 for the EsVan model and 0.824 for the EsVan2 model.

The investigators concluded: “Prospective, real-time clinical utilization of the EsVan models accurately predicts [bloodstream infection] risk and safely reduces unnecessary antibiotic use in febrile, nonseverely neutropenic pediatric patients with cancer.”

Adam J. Esbenshade, MD, MSCI, of Vanderbilt-Ingram Cancer Center, is the corresponding author for the Journal of Clinical Oncology article.

Disclosure: The study was funded by grants from the National Institutes of Health. For full disclosures of the study authors, visit

The content in this post has not been reviewed by the American Society of Clinical Oncology, Inc. (ASCO®) and does not necessarily reflect the ideas and opinions of ASCO®.