Prognostic Accuracy of BPD Definitions for Long-Term Outcomes in Preterm Infants: A Systematic Review

BACKGROUND AND OBJECTIVES Since the first description of bronchopulmonary dysplasia (BPD), multiple definitions to diagnose BPD and its grading have been published. Several studies have compared the predictive performance of these definitions for long-term outcomes. The objective was to identify the BPD definition with the optimal predictive performance for long-term respiratory and neurological outcomes in preterm infants. METHODS An electronic search identified studies in Medline and Embase from inception to August 2024. Studies assessing the performance of one or more BPD definitions for predicting long-term respiratory and/or neurological outcomes were included. We used the Quality in Prognostic Studies (QUIPS) tool for bias assessment. Reported prognostic accuracy of 5 BPD definitions (the 1988 Shennan, the 2001 National Institutes of Health NIH, the 2017 Canadian Neonatal Network, the 2018 NIH, and the 2019 Neonatal Research Network definition) was tabulated using specificity, sensitivity, C statistic, risk, or odds ratio. RESULTS Of the 6045 identified studies, 18 were included. Heterogeneity between studies resulted in inconsistent prognostic accuracy for long-term outcomes. The 2001 NIH definition showed higher prognostic accuracy for respiratory and neurological outcomes compared with the 1988 Shennan BPD definition. Only 5 studies showed a low to moderate risk of bias, and a sensitivity analysis confirmed the results. The limitations included challenges in comparing studies due to population heterogeneity and outcome definitions. CONCLUSIONS This systematic review shows that comparisons between the 2001 NIH definition and newer BPD definitions yield inconsistent results for predicting long-term outcomes. None of the current BPD definitions consistently provided sufficient prognostic accuracy for long-term respiratory and neurodevelopmental sequelae in very preterm infants.