Abstract
Background: Bronchopulmonary dysplasia (BPD) is the most common morbidity of very preterm (VPT) infants born <32 weeks’ gestation with life-long consequences. Studies document wide variation between regions and units in BPD prevalence.
Research question: Which unit-level factors contribute to the variation in BPD prevalence among very preterm infants between European neonatal units?
Study design and methods: Analyses were conducted using the prospective population-based EPICE cohort in 19 regions in 11 European countries. We compared prevalence of moderate/severe BPD among VPT infants without severe congenital anomalies in neonatal units with ≥40 annual VPT admissions (83 units, 5,285 infants). Unit prevalence was adjusted for individual risk factors using standardised morbidity rates. Spearman correlation and multilevel logistic regression were used to assess associations of BPD with unit-level variables: unit mortality rates, first week oxygen saturation targets, proportion of infants ventilated within the first 24 hours, unit practice of postnatal corticosteroid use for hypotension or BPD prevention and unit volume.
Results: Unadjusted BPD prevalence ranged from 2%-47% (median:13%) between units and was 8%-42% (median:17%) after adjustment and standardisation. Oxygen saturation targets, proportion of initial mechanical ventilation and postnatal corticosteroid use partly explained the between-unit variability (proportional change of variance: 25%, 5%, 17% respectively), leaving 53% unexplained. Risk-adjusted in-hospital mortality (range 8%-21%) and patient volume were not correlated with BPD prevalence.
Interpretation: Large variability in BPD prevalence exists between European units, which was only partially explained by patient characteristics. Our findings suggest that improving respiratory management for VPT infants could be beneficial for reducing BPD prevalence. The association of unit postnatal corticosteroid use practice with BPD requires further investigation.