Identification of factors predicting low-risk febrile neutropenia admissions in adults with acute myeloid leukemia

Febrile neutropenia (FN) is the most common reason for hospital readmission after chemotherapy for acute myeloid leukemia (AML) and is a major driver of health care resource utilization. Although FN risk models exist, they have largely been developed and validated for solid tumors. We therefore examined whether baseline characteristics could predict which patients with AML and FN have a lower risk of progression to severe illness. We identified adults with high-grade myeloid neoplasms (>= 10% blasts in the blood/marrow) who received intensive chemotherapy and who were admitted for FN between 2016 and 2023. We collected baseline clinical and disease variables. Outcomes were: infections identified, hospital length of stay (LOS), intensive care unit (ICU) admission, and survival. A lower-risk (LR) outcome was defined as LOS <72 hours without ICU admission or inpatient death. Univariate and multivariable (MV) logistic regression models were used to assess covariate associations with outcomes. We identified 397 FN admissions in 248 patients (median age, 61; [range, 29-77] years). The median hospital LOS was 6 days (range, 1-56) days; 10% required ICU admission, and 3.5% died inpatient. Only 15% of admissions were LR. Infection was identified in 59% of admissions. Physiologic parameters, including heart rate, blood pressure, and fever height, were the best predictors of LR admission and infection. We developed MV models to predict LR admission and infection with area under the curve (AUC) of 0.82 and 0.72, respectively. Established FN and critical illness models were not predictive of outcomes in AML, and we could not identify a LR group; thus, an AML-specific FN risk model requires further development and validation.