To assess associations between neonatal intensive care unit (NICU)-level patent ductus arteriosus (PDA) treatment rates (pharmacologic or surgical) and neonatal outcomes.
This cohort study included infants born at 24-28 weeks of gestation and birth weight <1500 g in 2007-2015 in NICUs caring for ≥100 eligible infants in 6 countries. The ratio of observed/expected (O/E) PDA treatment rates was derived for each NICU by estimating the expected rate using a logistic regression model adjusted for potential confounders and network. The primary composite outcome was death or severe neurologic injury (grades III-IV intraventricular hemorrhage or periventricular leukomalacia). The associations between the NICU-level O/E PDA treatment ratio and neonatal outcomes were assessed using linear regression analyses including a quadratic effect (a square term) of the O/E PDA treatment ratio.
From 139 NICUs, 39 096 infants were included. The overall PDA treatment rate was 45% in the cohort (13%-77% by NICU) and the O/E PDA treatment ratio ranged from 0.30 to 2.14. The relationship between the O/E PDA treatment ratio and primary composite outcome was U-shaped, with the nadir at a ratio of 1.13 and a significant quadratic effect (P<.001). U-shaped relationships were also identified with death, severe neurologic injury, and necrotizing enterocolitis.
Both low and high PDA treatment rates were associated with death or severe neurologic injury, whereas a moderate approach was associated with optimal outcomes.