363 - Bronchopulmonary Dysplasia Definition Based on Respiratory Severity Score

Publication Number: 1347.363

Dinushan C. Kaluarachchi, University of Wisconsin - Madison, Madison, WI, United States; Patrick J. Peebles, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States; Michael R.. Lasarev, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States; Scott Guthrie, Children's Hospital at Erlanger, Chattanooga, TN, United States; Matthew M. Laughon, The University of North Carolina at Chapel Hill, Chapel Hill, NC, United States

Background: Bronchopulmonary Dysplasia (BPD) grade is currently defined based on the type of respiratory support at 36 weeks postmenstrual age (PMA). A definition that focuses only on respiratory support type may have limitations, given it doesn't include mean airway pressure (MAP) or the fraction of oxygen (FiO2). In this study, we evaluate a novel BPD grading system based on the respiratory severity score (RSS), which accounts for both MAP and FiO2.

Objective: Evaluate the performance of a BPD definition based on RSS (MAP x FiO2) at 36 weeks PMA in predicting severe respiratory morbidity severity (RMS) compared to the BPD definition by Jensen et al.

Design/Methods: This is a retrospective cohort study of preterm infants born at 23 0/7 to 28 6/7 weeks' gestation from the Prematurity and Respiratory Outcomes Program. Deaths or transfers before 36 weeks PMA and infants with inadequate data to calculate RSS or RMS were excluded. RMS was defined according to the original study criteria (Table 1). BPD grading was defined according to two BPD definitions: 1) the definition by Jensen et al based on type of respiratory support at 36 0/7 weeks; 2) the proposed BPD definition based on average RSS at 36 weeks PMA (Table 1). Each patient's daily RSS was for the week of 36 weeks' PMA (36 0/7 to 36 6/7) and the average RSS for 36 weeks' PMA was calculated. Non-zero average RSS values at 36 weeks' PMA were divided into three tertiles to create three BPD grades. Log-binomial regression was used to determine whether risk of severe RMS (Table 1) was associated with severity of BPD according to Jensen definition or with the RSS based BPD definition. Concordance statistics (C-index) were calculated to assist in evaluating how successful each definition was in predicting severe RMS.

Results: The study included 683 infants. A total of 293 (43%) were on room air throughout the week of 36 weeks' PMA and were assigned an RSS of 0 by both definitions. The three RSS tertiles were grouped as follows: grade 1 (RSS 0.01-0.25), grade 2 (RSS 0.26-0.64), and grade 3 (RSS >=0.65) (Figure 1). Log-binomial regression revealed that risk of severe RMS was considerably elevated in each non-zero level of BPD for both definitions (p < 0.001 for each risk ratio) (Table 2). C -indices treating each definition as an ordered factor, had the same ability to predict infants with severe RMS (C = 0.67; 95% CI: 0.63-0.71 for each) with no difference noted between the two definitions (p=0.89).

Conclusion(s): There was a strong correlation between the two BPD grading systems. There was no difference in performance between the two BPD definitions in predicting severe RMS.

Table 1. Definitions for Bronchopulmonary dysplasia (BPD), Respiratory morbidity severity (RMS) and respiratory severity score


Table 2. Estimated risk ratios (RR) and 95% confidence intervals (CI) for severe RMS based on BPD grades according to definition by Jensen et al and proposed BPD definition based on average RSS during 36 weeks' PMA.
Concordance measures are reported for each model. P-values for all risk ratios are <0.001, but no performance difference noted between models using different definitions (p=0.891).

Figure 1. Distribution of average respiratory severity score during 36 weeks' post menstrual age (n = 683).
Individual points are colored according to the BPD definition by Jensen et al. Overlayed boxplot reveals the median (thick black line) at 0.33 with corresponding IQR as box edges (at 0.25 & 0.83). Tertile limits (33rd & 67th percentiles) are drawn along the right edge and placed at 0.25 and 0.64, respectively.