Analysis of predictive parameters for extubation in very low birth weight preterm infants

Background: Mechanical ventilation is the primary treatment for preterm infants with respiratory failure. Prolonged intubation may lead to complications; thus, early extubation is desirable. No standard criteria exist for determining the appropriateness of extubating very-low-birth-weight (VLBW) infants. This study explored the predictors of successful extubation in preterm VLBW infants. Methods: This retrospective cohort study included 60 preterm VLBW infants who underwent their first extubation in the neonatal intensive care unit in a regional hospital in Hsinchu, Taiwan, between January 2017 and November 2020. Successful extubation was defined as having no requirement of reintubation within 3 days of extubation. Potentially predictive variables, including demographics, prenatal characteristics, and ventilator parameters were compared between a successful extubation group and failed extubation group. Results: Of the 60 infants, 47 (78.33%) underwent successful extubation. The successful extubation group had higher Apgar scores at 1 (7 vs. 6, P = 0.02) and 5 min (9 vs. 7, P = 0.007) than those of the failed extubation group. Ventilator inspiratory pressure and mean airway pressure were significantly lower at 24, 16, 8, and 1 h before extubation and upon its completion in the successful extubation group. The areas under a number of the receiver operating characteristic curve curves in this study were moderate, specifically, 0.72, 0.74, and 0.69. Statistical analysis revealed an association between ventilator parameters before 1 h extubation (IP > 17.5cmH₂O, MAP >7.5 cmH₂O, RSS >1.82) and extubation failure (odds ratio 1.73, 2.27, 2.46 and 95% confidence interval:1.16–2.6, 1.26–4.08, 1.06–5.68, respectively). Conclusion: Higher Apgar scores at birth, lower ventilator inspiratory pressure, and mean airway pressure 24, 16, 8, and 1 h and 1 h RSS prior to extubation are associated with successful extubation in VLBW preterm infants.