Objectives: High-flow nasal cannula (HFNC) use in bronchiolitis has increased substantially with wide variability, likely due to the lack of objective means of predicting clinical outcomes. Our objective was to identify features associated with care escalation for infants started on HFNC in the Emergency Department (ED), thereby assisting in disposition of patients and optimizing the utilization of limited resources.
Methods: This is a retrospective cohort study from 3 free-standing children's hospitals of infants younger than or equal to 12 months with bronchiolitis who were initiated on HFNC in the ED between 2/1/2018 and 3/1/2020. The primary outcome was escalation of care within the first 24 hours after HFNC initiation [transfer to pediatric intensive care unit (PICU) and/or initiation of noninvasive positive pressure ventilation (NIPPV) or mechanical ventilation (MV)]. A clinical prediction model was developed using multivariable logistic regression.
Results: We included 738 infants with a mean age of 5.4 months (SD 3.4), of which 73 (10%) experienced care escalation within the first 24 hours. Fever, tachypnea, tachycardia, and hypoxemia were associated with escalation, whereas historical features, physical examination findings, and response to ED therapies were not. A prediction model consisting of the highest ED respiratory rate and the lowest oxygen saturation yielded an AUC of 0.75, with a sensitivity of 57.9% and specificity of 82.2% (based on an optimal predicted probability threshold of 7.9%).
Conclusions: Hypoxemia and tachypnea predicted early escalation of care in infants with bronchiolitis with moderate performance. This model may have promise to assist with management decisions; however, it requires prospective validation.
Keywords: bronchiolitis; high flow nasal cannula; respiratory failure; statistical modeling.
Copyright © 2026 Wolters Kluwer Health, Inc. All rights reserved.