Prospective External Validation of the Esbenshade Vanderbilt Models Accurately Predicts Bloodstream Infection Risk in...

Area of Science:

• Pediatric Oncology
• Infectious Diseases
• Clinical Decision Support

Background:

• Fever without severe neutropenia in pediatric cancer patients lacks defined optimal management.
• Existing Esbenshade Vanderbilt (EsVan) models predict bacterial bloodstream infections (BSIs) but require external validation.

Purpose of the Study:

• To prospectively validate the EsVan models for predicting BSIs in pediatric cancer patients with fever and non-severe neutropenia.
• To assess the models' ability to stratify risk and guide management decisions.

Main Methods:

• Prospective data collection from 2,565 episodes in 18 academic medical centers.
• Application of five EsVan model versions to predict overall BSI and high-risk organism BSI.
• Evaluation of model performance using C-statistics and calibration.

Main Results:

• The BSI rate was 4.7%, with low rates of ICU admission (1.1%) and mortality (0.2%).
• EsVan models demonstrated strong predictive accuracy (C-statistics 0.775-0.819) for BSIs, particularly high-risk organisms.
• Withholding empiric antibiotics in 14.9% of episodes was safe, with no attributable deaths or ICU admissions.

Conclusions:

• The EsVan models, particularly EsVan2b, show excellent prospective performance in risk-stratifying non-neutropenic fever in pediatric cancer.
• Routine screening with risk-stratified management can safely decrease unnecessary antibiotic use in this population.