# Development and Validation of a Prognostic Risk Score for Patients With Cancer and Neutropenic Fever Presenting to the Emergency Department

## TL;DR

This study creates a new risk score to predict outcomes for cancer patients with neutropenic fever in emergency departments, aiming to improve care and reduce hospital stays.

## Contribution

A novel, objective, and externally validated risk score for neutropenic fever patients in the emergency department is developed.

## Key findings

- The risk score demonstrated good predictive accuracy with an AUC of 0.77 in derivation and 0.78 in validation cohorts.
- Low-risk patients had 0% in-hospital mortality and less than 2% ICU admission rates.
- The score uses readily available clinical and lab data, supporting real-time implementation.

## Abstract

Neutropenic fever (NF) is a life-threatening oncologic emergency associated with significant morbidity and mortality. Current predictive risk-stratification scores’ limitations include outdated criteria, subjectivity, and low usage in the emergency department (ED). We sought to develop and externally validate a pragmatic, objective risk score to predict adverse outcomes in patients presenting to the ED with cancer and NF.

A retrospective cohort study was performed using data from 2 comprehensive cancer centers. The derivation cohort included adults presenting with NF to Memorial Sloan Kettering Cancer Center. External validation was conducted using data from the MD Anderson Cancer Center. The primary composite adverse outcome included intensive care unit, bacteremia, oxygen supplementation therapy, hospital stay >3 days, and in-hospital mortality. Predictors were assigned points based on the final model coefficients.

The derivation and validation cohorts included 827 and 777 patients, respectively. The final score that included multiple clinical and laboratory biomarkers demonstrated an area under the curve of the receiver operating characteristic curve value of 0.77 (95% CI, 0.73-0.80) in the derivation cohort and 0.78 (95% CI, 0.74-0.83) in the validation cohort. In patients classified as low risk, in-hospital mortality was 0.0%, and intensive care unit admission occurred in < 2% of cases across both cohorts.

The newly developed risk score effectively stratified patients with NF in the ED. Its reliance on readily available, non-subjective data supports the feasibility of its real-time implementation and could improve decision-making and reduce unnecessary hospitalization. Prospective validation in diverse clinical settings is needed.