P0708 - Assessing 3 Bronchopulmonary Dysplasia Definitions: Associations between Room Air Challenge Results and Respiratory Outcomes

Background:
The Neonatal Research Network (NRN) has used 3 definitions for bronchopulmonary dysplasia (BPD): 1) a clinical definition based on supplemental oxygen at 36 weeks postmenstrual age (PMA) (clinBPD), 2) a physiologic definition inclusive of a room air challenge (RAC) at 36 weeks (physBPD), and 3) a pragmatic definition based on any respiratory support at 36 weeks (pragBPD). PragBPD has been shown to be strongly correlated with long term respiratory outcomes.

Objectives: Evaluate variations in prevalence of the 3 BPD definitions at NRN centers and over time, and compare outcomes for infants that pass or fail the RAC.

Design/

Methods:
Cohort study of infants 30% supplemental oxygen, or 2) ≤ 30% oxygen via hood or nasal cannula with failure of the RAC (oxygen saturation <90% for 5 minutes, or <80% for 15 consecutive seconds). PragBPD was defined as any respiratory support at 36 weeks PMA categorized into 3 severity grades. We compared prevalence of the BPD types over time and by center. We compared demographics and clinical characteristics among infants with the 3 BPD types, and infants who passed or failed the RAC. For infants <27 weeks GA, we examined associations between RAC results and death, long-term respiratory morbidity, or moderate-severe neurodevelopmental impairment at 22-26 months PMA.

Results:
Of 8026 infants included, 1074 completed the RAC and 554 of these completed 22-26 month follow-up. The prevalence of each definition of BPD has significantly increased over time. In pairwise comparisons between the 3 BPD definitions, Kappa coefficients ranged from 0.50-0.88 among baseline characteristics (Table 1) and 0.52-0.97 among centers. Demographic and clinical characteristics were not significantly different between infants who passed or failed the RAC, other than receipt of steroids (Table 1). The RAC results were not associated with long-term morbidities (Table 2).

Conclusion:
Prevalence of all 3 BPD definitions increased over time; pragBPD was higher in every center assessed. Because RAC results were not predictive of long-term morbidities, pragBPD may be a more useful BPD definition for trials and quality improvement benchmarking.