Date of Award

January 2026

Document Type

Thesis

Degree Name

Medical Doctor (MD)

Department

Medicine

First Advisor

John Giuliano

Second Advisor

Michael Canarie

Abstract

BACKGROUND: Acute respiratory failure (ARF) is a leading cause of admission to the pediatric intensive care unit (PICU) and a major contributor to morbidity in hospitalized children. While noninvasive ventilation (NIV) is increasingly used to avoid the complications of endotracheal intubation (ETI) and invasive mechanical ventilation (IMV), NIV failure remains common. Delayed recognition of NIV failure may result in prolonged hypoxemia, hemodynamic instability, and emergent intubation. Despite widespread use of NIV in contemporary pediatric practice, predictors of NIV failure have not been consistently defined in modern cohorts.

The aim of this study was to identify clinical variables associated with progression from NIV to IMV in children admitted with ARF. We hypothesized that escalating ventilatory support parameters and worsening oxygenation indices during NIV would be associated with increased risk of intubation.

METHODS: We conducted a retrospective observational study of children under 18 years of age admitted to the Yale New Haven Children’s Hospital PICU between January 2020 and December 2023 with ARF treated with NIV. Patients intubated within two hours of NIV initiation were excluded to focus on clinically meaningful NIV trials. Demographic characteristics, principal diagnoses, ventilatory settings, and oxygenation parameters were abstracted from the electronic medical record. Primary predictors included maximum inspiratory positive airway pressure (IPAP), maximum expiratory positive airway pressure (EPAP), maximum fraction of inspired oxygen (FiO₂), and lowest saturation-to-FiO₂ (S:F) ratio. The primary outcome was progression to invasive mechanical ventilation. Discrimination was assessed using receiver operating characteristic (ROC) curves, calibration was evaluated using the Hosmer–Lemeshow test, and optimal thresholds were determined using the Youden index.

RESULTS: Among 411 children treated with NIV, 54 (13.1%) required intubation. Maximum EPAP demonstrated the strongest individual discrimination for NIV failure (AUC 0.75), followed by maximum IPAP (AUC 0.71), lowest S:F ratio (AUC 0.69), and maximum FiO₂ (AUC 0.68). A multivariable model incorporating IPAP, EPAP, and FiO₂ achieved an AUC of 0.78 but did not significantly outperform EPAP alone. An EPAP threshold of 8 cmH₂O yielded a sensitivity of 75%, specificity of 62%, positive predictive value of 26%, and negative predictive value of 93%.

CONCLUSION: In this contemporary PICU cohort, escalating EPAP during NIV was associated with progression to invasive ventilation and performed comparably to more complex multivariable models. These findings suggest that routinely available ventilatory parameters may support earlier recognition of children at increased risk for NIV failure. Improved risk stratification has the potential to inform bedside reassessment, enhance communication with families regarding trajectory, and reduce variability in escalation practices. Prospective multicenter validation is warranted to determine generalizability and guide integration into clinical decision-support strategies.

Comments

This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.

Download

Share

COinS