Document resource
Objectives To examine bronchopulmonary dysplasia (BPD) epidemiological data in Chinese very preterm infants and compare agreement between four diagnostic definitions and their predictive accuracy for discharge outcomes.Design Observational epidemiologic study of a multicentre cohort of the Chinese Neonatal Network (CHNN).Setting Tertiary neonatal intensive care units participating in the CHNN.Participants 42 664 preterm infants born at <32 weeks’ gestational age from 2019 to 2023.Measures BPD was diagnosed using four definitions: Shennan 1988, the National Institute of Child Health and Human Development (NICHD) 2001 and 2018, and the Neonatal Research Network (NRN) 2019 definitions. BPD prevalence and severity were examined. Agreement was assessed using kappa statistics. Predictive accuracy for discharge outcomes was evaluated using c-statistics from multivariable generalised estimating equation models.Results Among 42 664 infants (mean gestational age, 29.0 weeks; 43.1% females), BPD prevalence varied significantly: Shennan 1988: 37.0%, NICHD 2001: 51.1%, NICHD 2018: 37.0%, NRN 2019: 37.0%. NICHD 2001 and 2018 definitions classified more infants as severe BPD (16.4% and 10.1%, respectively), while NRN 2019 classified more as grade 2 (moderate; 15.0%). Shennan 1988 showed good agreement with NICHD 2018 (κ=1.0) and Jensen 2019 (κ=1.0). Shennan 1988 (c-statistic range: 0.921–0.974), NICHD 2018 (0.948-0.978) and NRN 2019 (0.949-0.982) demonstrated higher discriminative accuracy for discharge outcomes than NICHD 2001 (0.854–0.925).Conclusions This study found a high prevalence of BPD among very preterm infants in China, varying by definitions. The Shennan 1988, NICHD 2018 and NRN 2019 definitions showed good agreement and better predictive accuracy for outcomes at discharge compared with NICHD 2001. These definitions could be prioritised for clinical use in our population.