382 - Association of Carbon Dioxide Levels to Severity of Bronchopulmonary Dysplasia and In-Hospital Clinical Outcomes in Premature Infants.

Publication Number: 1366.382

Susan McAnany, Children's Mercy Hospitals and Clinics, Overland Park, KS, United States; Mary Decker, University of Missouri - Kansas City, Kansas City, MO, United States; Keith Feldman, University of Michigan Medical School, Ann Arbor, MI, United States; Venkatesh Sampath, Children's Mercy Kansas City, kansas city, KS, United States

Background: Bronchopulmonary dysplasia (BPD) represents a common complication of prematurity yet assessment remains imprecise. Commonly used BPD definitions only consider respiratory support and oxygen requirement at 36 weeks' postmenstrual age (PMA), failing to capture nuance on infants physiologic conditions. Recent studies suggest infants with BPD often have higher than normal carbon dioxide (CO2) levels secondary to their lung pathology, suggesting blood gases may offer utility of consideration of BPD related morbidities.

Objective: To determine whether CO2 levels; a) correlate with the severity of BPD as defined by Neonatal research network (NRN) criteria, and b) predict time (in days) to wean to low flow nasal cannula (LFNC) and days to discharge.

Design/Methods: Retrospective cohort study of all infants with BPD admitted to Children's Mercy Hospital NICU or Kansas University Medical Center NICU between Jan 1, 2018 - Dec 31, 2023. Infants with severe congenital anomalies or were admitted ≥ 7 days of life were excluded. For each infant, all CO2 levels were collected from blood gases taken between 34 and 36 weeks PMA. Relationship to NRN BPD levels were calculated with a Jonckheere-Terpstra trend test. A linear mixed effect model was then used to assess the relationship of CO2 to time to two key clinical outcomes, days to LFNC and days to discharge, adjusting for birth weight, gestational age, and infant sex, with a random effect for clinical center. For completeness, models were run for both maximum (max) and median lab values independently.

Results: We obtained a total of 473 CO2 values for the 81 infants in this study. Infants with more severe disease (NRN grade 3) had more lab values compared to infants with less severe disease (NRN grades 1 or 2). Infants with more severe disease have higher CO2 values (Figures 2A and 2B) with average max CO2 levels being 56.7, 62.0, and 78.9 and average median CO2 levels being 52.8, 55.1, and 57.2 for NRN grades 1, 2, and 3 respectively (Max p-value < < 0.001, Median: 0.075). We identified similar trends between the two sites. Infants with higher CO2 levels took longer to wean to LFNC and to be discharged (Table 3).

Conclusion(s): This pilot study suggests that CO2 levels correlate with severity of BPD and are meaningful associated with in-hospital outcomes, such as days to LFNC. This study is limited by a small sample size and therefore a large, multi-center cohort study is needed. Future studies are needed to determine how the utilization of CO2 values, in addition to respiratory support and oxygen requirements, can better predict BPD severity and lead to improved counseling.

Table 1A: Demographic data and outcome of the study cohort; Table 1B: Total and average number of CO2 levels collected per NRN grade of BPD severity
Table 1A: BPD, bronchopulmonary dysplasia; NRN, neonatal research network; LFNC, low flow nasal cannula.

Table 1B: Infants with more severe disease had more CO2 levels (obtained from blood gases) than infants with less severe disease. NRN, neonatal research network; BPD, bronchopulmonary dysplasia; CO2, carbon dioxide.

Figure 2A: Box plot showing maximum CO2 levels by grade of BPD based on NRN Criteria; Figure 2B: Box plot showing median CO2 levels by grade of BPD based on NRN Criteria
Figure 2A: Average maximum CO2 levels based on NRN Grades 1, 2, and 3 are 56.7, 62.0, and 78.9. P-value of the trend between grades was < < 0.001 (obtained using Jonckheere-Terpstra test). CO2, carbon dioxide; BPD, bronchopulmonary dysplasia; NRN, neonatal research network.

Figure 2B: Average median CO2 levels based on NRN Grades 1, 2, and 3 are 52.8, 55.1, and 57.2. P-value of the trend between grades was ~ 0.075 (obtained using Jonckheere-Terpstra test). CO2, carbon dioxide; BPD, bronchopulmonary dysplasia; NRN, neonatal research network.

Table 3: Regression table representing days to LFNC and days to discharge based on median and max CO2 levels
Infants with higher CO2 levels (using either median CO2 levels or maximum CO2 levels) took longer to wean to LFNC and took longer to discharge. LFNC, low flow nasal cannula; CO2, carbon dioxide.