Diagnostic Classification of Bronchopulmonary Dysplasia: A Compromise between Defining Lung Disease versus Long-Term Outcome Prediction.

To the Editor:

There are important limitations to current bronchopulmonary dysplasia (BPD) definitions, and many groups are working to come up with diagnostic criteria that are better adapted to current clinical presentation and treatment modalities and can also predict long-term outcomes.

Jensen and colleagues analyzed data from the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network to explore 18 different combinations of respiratory support and identify the definition that best predicts death or poor long-term respiratory and neurological outcomes (1). The authors concluded that a definition that includes nasal cannula flow, nasal continuous positive airway pressure, and mechanical ventilation at 36 weeks corrected age offers the best prediction for these composite outcomes. Surprisingly, many of the combinations of respiratory support tested in this study showed very similar accuracy for predicting long-term outcomes.

Although the proposed definition is appealing because of its simplicity and ability to predict outcomes and was validated in a large, multicenter population, it may not accurately reflect the severity of lung disease. This analysis assumes that the use of respiratory support in preterm infants is driven mainly by parenchymal lung disease. In reality, the use of respiratory support in this population can be related to many different respiratory and nonrespiratory problems, and indications for such support are subjective and vary considerably among institutions. Therefore, many patients may be inappropriately labeled as having BPD when in fact they are receiving respiratory support for indications other than lung disease.

More surprisingly, and in contrast to previous evidence (2–4), the authors concluded that inspired oxygen administered at 36 weeks postmenstrual age did not add strength to the prediction models. Oxygen administration may vary among centers, but in most instances, oxygen is titrated to maintain a narrow range of SaO. In the absence of extrapulmonary shunts, inspired oxygen is the simplest and most sensitive single indicator of the severity of respiratory failure and parenchymal lung disease. Few clinicians would question the fact that an infant who requires 50% oxygen has worse lung function than one who is receiving 25% oxygen to maintain normal arterial oxygenation. Before concluding that the need for oxygen at 36 weeks postmenstrual age is not informative about the severity of lung disease or long-term morbidities, we need to thoroughly evaluate the association between different fractions of inspired oxygen and long-term respiratory or neurologic outcomes while adjusting for other confounding factors.

Another limitation of the proposed definition is that it is based on a one-time assessment. In this situation, a preterm infant who exhibits acute respiratory deterioration at approximately 36 weeks and needs mechanical ventilation would be labeled as having severe BPD even if his or her lungs were normal. This limitation can be easily avoided by including an indicator of the chronicity of the parenchymal lung disease. Although the proposed definition may work in a large population base and predict long-term outcomes, it will not always reflect the severity of lung disease and therefore will not be appropriate for research or benchmarking focused on pulmonary outcomes.

Although most clinicians would agree that more contemporary definitions of BPD are necessary, these definitions must be based on the premise that BPD is secondary to chronic parenchymal lung disease. Therefore, any definition of BPD, whether based on diagnostic or therapeutic criteria, should reflect the severity of lung disease and exclude nonpulmonary indications for respiratory support. The proposed definition, though marginally better at predicting outcomes, will not consistently meet the challenge of defining BPD.