Objective To investigate whether the wide variation in the frequency of bronchopulmonary dysplasia (BPD) between hospitals is due to differences in delivery room intubation rates.

Methods Data on 1260 infants of birth weight <1500 g and 23–31 weeks gestational age, born in 1999–2002 and surviving to 36 weeks, were collected; 196 (15.6%) developed BPD defined as oxygen need at 36 weeks postmenstrual age. Generalised estimating equations and conditional logistic models adjusting for centre, gestational age, propensity score for intubation, and other potential confounders were used.

Results Rates of BPD, delivery room intubation and mechanical ventilation for >24 h differed significantly between hospitals. Centres with high delivery room intubation rates had higher ventilation and BPD rates. Hospitals ventilating more often also did so for a longer time. Although delivery room intubation was associated with BPD in unadjusted analyses, neither delivery room intubation nor brief (<24 h) mechanical ventilation were risk factors for BPD in multivariate analyses adjusting for gestational age, case mix and other pre- and perinatal factors, indicating no causal effect or unmeasured confounding. Significant risk factors for developing BPD were low gestational age, prolonged ventilation (>24 h: adjusted OR (aOR) 2.4; >7 days: aOR 14.9), male sex (aOR 1.7), being small for gestational age (SGA; aOR 4.3) and late-onset sepsis (aOR 2.2). After taking into account these variables/procedures, centre differences remained significant but explained only about 5% of variance.

Conclusions Differences in BPD frequency between hospitals are explained by differences in procedures, chiefly mechanical ventilation, rather than by differences in initial management or case mix. Delivery room intubation and brief mechanical ventilation did not increase BPD risk.