WIP 70 - Initial Mode of Invasive Ventilation in an Extreme Premature Infant Cohort: First-intention High Frequency Jet Ventilation, Conventional Ventilation Failure and Related Predictors and Outcomes.

Poster Number: WIP 70
Publication Number: WIP 70.800

Background: Bronchopulmonary dysplasia (BPD) leads to mortality and morbidity in premature infants. Mechanical ventilation is related to BPD via biotrauma and barotrauma. High Frequency Jet Ventilation (HFJV) is a gentle mode of ventilation with theoretical less atelectotrauma and has been adopted as initial mode of ventilation in extreme premature infants. Data remains equivocal with dated RCTs and limited cohort studies. We hypothesize that HFJV improves outcomes in extremely premature infants. We will evaluate predictors of conventional ventilation failure to identify patients who may benefit from initial HFJV.

Objective: Describe outcomes related to first-intention HFJV in premature infants compared to conventional ventilation or its failure.
Determine predictors such as neonatal or perinatal characteristics or biomarkers of failure of conventional ventilation.

Design/Methods: Retrospective cohort study of infants born between 2009 -2019 under 29 weeks of gestation requiring invasive ventilation. Approved by Institutional Review Board (Protocol: NA_00026068). Groups are: Infants on HFJV as the initial mode, those on conventional ventilation only, and those who failed conventional ventilation. Exclusion criteria are diagnoses of congenital pulmonary/heart anomalies or incomplete biomarker data. IL6, IL8, IL10 and VEGF serum levels were obtained through 7 days. Outcomes are mortality, time to extubation, BPD severity, and duration of admission.
Cohort characteristics and biomarkers will be compared by group using appropriate non-parametric tests. Outcomes will be assessed using logistic regression with the conventional ventilation group as reference. Sensitivity analyses for outcomes include Kaplan-Meier analyses for time to death, time to extubation. Confounders such as birth weight and gestational age will be adjusted for. P values < 0.05 will be considered significant. Analyses are conducted using STATA 18 (College Station, TX). Data is obtained (n=229) and preliminary analysis have been completed. Regression analyses and model fitting will be completed by 01/2024.