Need for an International Consensus on the Definition of Bronchopulmonary Dysplasia.

To the Editor:

The definition and diagnosis of bronchopulmonary dysplasia (BPD) are widely debated topics in neonatology (1). The half-century-long history of different definitions highlights the changing paradigm for managing pulmonary immaturity (2). Jensen and colleagues compared 18 BPD definitions based on levels of supplemental oxygen and respiratory support provided at 36 weeks postmenstrual age (PMA) (3). They reported categorizations of BPD severity that predicted serious respiratory morbidities or mortality at 18–26 months corrected age. This study is a useful addition to the literature; however, it inherently accepts a fundamental misconception that 36 weeks PMA is somehow a “magic” postnatal age when preterm infants are mature enough to be off respiratory support. There is no physiological or clinical basis for using 36 weeks PMA as a cutoff.

The authors assessed various categorizations of BPD severity only at 36 weeks PMA and did not consider other PMA timings because of data limitations. It is well known that BPD at 36 weeks PMA is not a robust marker for predicting long-term outcomes (1). Physiological and developmental considerations have led investigators to suggest using a PMA closer to the expected due date, because even late-preterm and early-term infants of 34–38 weeks gestation have a higher risk of developing respiratory issues than infants of 39–41 weeks gestation (4). Although using a later PMA may increase missed assessments due to transfer of infants to step-down units or discharge home before assessment, obtaining such information from step-down units or families is not impossible. Shennan and colleagues chose 36 weeks PMA as a cutoff for BPD based on a compromise between sensitivity and specificity; however, in the 30 years since their work was published, there have been marked changes in the relevant population (with gestational ages as low as 22 wk), approaches, and technology available to provide respiratory support (both invasive and noninvasive) (5). Thus, applying a cutoff without any physiological basis seems to represent a halt in progress. Moreover, the first 12 BPD definitions assessed by Jensen and colleagues had similar predictive abilities, with a difference in area-under-the-curve values of 0.01. A recent study reported that the ability to predict long-term respiratory morbidities improved as the timing to define or diagnose BPD increased from 36 to 40 weeks PMA, indicating that 40 weeks PMA is the best time for predicting respiratory outcomes (1, 6). Evaluation of the data from Jensen and colleagues may provide supporting or refuting evidence for this maturational cutoff for defining BPD.

We agree that our goal should be to adopt an evidence-informed or data-driven approach to identify neonates of extremely low gestational age with a future risk of developing pulmonary and neurodevelopmental issues. The critical task for members of the neonatal community is to decide what we aim to achieve in defining BPD. Do we want to identify nearly all children who may develop adverse outcomes (minimum false negatives), do we want to rule out all who may not develop adverse outcomes (minimum false positives), or do we want to settle for a compromise and accept a middle ground? The answer may require careful thinking. We may want to use criteria with minimum false negatives when identifying children for closer surveillance during childhood, to predict and manage respiratory adverse outcomes; use criteria with minimum false positives when testing experimental therapies, to rationalize exposure for many children; and use “compromise” criteria (with acceptable sensitivity and specificity cutoffs) for quality improvement initiatives, benchmarking, and assessing trends. Discussions about these issues need to happen through an international forum and consensus process, as is currently underway via the International Neonatal Consortium (1). Purpose-defined BPD criteria derived from an international consensus process and supported by data are essential for avoiding ongoing confusion and inconsistency.