Abstract Type

Research

Primary Mentor

Keith Feldman, Children's Mercy Hospital (kfeldman@cmh.edu)

Start Date

14-5-2021 11:30 AM

End Date

14-5-2021 1:30 PM

Presentation Type

Poster Presentation

Description

Background: Despite shifts to non-invasive support, as of 2012, 87% of preterm infants admitted to the neonatal intensive care unit (NICU) continued to receive some form of invasive mechanical ventilation1 . As a result, research around how prolonged respiratory support may impact the development of chronic lung disease, specifically bronchopulmonary dysplasia (BPD) has flourished. To date, literature has established associations between the time on invasive mechanical ventilation and incidence BPD2,3. Resulting in the development of standardized extubation strategies and bringing with them objective measures of extubation readiness4,5. Yet, due to their volatile conditions, some neonates will inevitably need to be reintubated. Given the established risk of invasive ventilation, the decision of when to do so, varies widely between care teams; integrating patient’s physiologic state and physician’s experience. Stemming from this lack of standardization, patient conditions vary widely before reintubation, and it remains unclear if the degree of deterioration in infant’s physiologic state during periods of non-invasive ventilatory support (NIV) may also be associated with downstream lung injury. Quantifying such associations may serve to guide evidence-based practice for future reintubation guidelines.

Objective: The objective of this study was to determine the association between variability in premature infants’ physiologic state prior to reintubation and BPD classification at 36-weeks. A secondary objective aimed to identify empirical deterioration thresholds to provide risk-based guidance for reintubation.

Data & Analysis: Data were retrospectively drawn from the Children's Hospitals Neonatal Database (CHND) and electronic medical records (EMR) of infants admitted to Children’s Mercy NICU 2010-2016. Inclusion criteria required infants to be preterm, admitted within three days of life, and intubated within one day post-admission. Infants must also have been intentionally extubated and subsequently reintubated prior to 36-weeks, those with multiple extubations before 36-weeks were excluded. BPD status at 36-weeks was classified per the CHND; and grouped into (None/MildBPD) and (Severe-BPD/Death) for analysis. Automated methodology for extracting periods of NIV support from EMR data was developed and validated on a random 30 infants (~5%). For each infant, timestamped data was extracted, including measures of respiratory support (FiO2) and physiologic measurements (blood gases, respiratory rate, SpO2). Bayesian logistic regression was used to estimate the magnitude and directionality of association between NIV measurements and BPD outcomes. Models were adjusted for known confounders, including demographics (gestational age, birthweight, sex, race), use of surfactant or antenatal steroids, and timing (extubation day of life, hours extubated, and time reintubated before assessment).

Results: All infants with complete data were utilized, resulting in a cohort of 124. Beyond known relationships to BPD (e.g., sex, length of time reintubated), we identified novel associations between the maximal and minimal values of blood gas measures during NIV, and BPD Outcomes. Complete results can be found in Table 1.

Conclusions: This work offers a first step to understanding how patient deterioration during NIV may be associated with longerterm outcomes. Future work includes more comprehensive characterization of NIV blood gas trajectories, and exploration into high-risk thresholds. Understanding these relationships may aid in standardization of reintubation strategies to improve patient morbidity.

MeSH Keywords

Intensive Care Units, Neonatal; Data Science; Ventilation

Share

COinS
 
May 14th, 11:30 AM May 14th, 1:30 PM

Variability in Blood Gas Levels During Non-Invasive Ventilatory Support Following Planned Extubation and Association to 36-Week Bronchopulmonary Dysplasia in Preterm Neonates

Background: Despite shifts to non-invasive support, as of 2012, 87% of preterm infants admitted to the neonatal intensive care unit (NICU) continued to receive some form of invasive mechanical ventilation1 . As a result, research around how prolonged respiratory support may impact the development of chronic lung disease, specifically bronchopulmonary dysplasia (BPD) has flourished. To date, literature has established associations between the time on invasive mechanical ventilation and incidence BPD2,3. Resulting in the development of standardized extubation strategies and bringing with them objective measures of extubation readiness4,5. Yet, due to their volatile conditions, some neonates will inevitably need to be reintubated. Given the established risk of invasive ventilation, the decision of when to do so, varies widely between care teams; integrating patient’s physiologic state and physician’s experience. Stemming from this lack of standardization, patient conditions vary widely before reintubation, and it remains unclear if the degree of deterioration in infant’s physiologic state during periods of non-invasive ventilatory support (NIV) may also be associated with downstream lung injury. Quantifying such associations may serve to guide evidence-based practice for future reintubation guidelines.

Objective: The objective of this study was to determine the association between variability in premature infants’ physiologic state prior to reintubation and BPD classification at 36-weeks. A secondary objective aimed to identify empirical deterioration thresholds to provide risk-based guidance for reintubation.

Data & Analysis: Data were retrospectively drawn from the Children's Hospitals Neonatal Database (CHND) and electronic medical records (EMR) of infants admitted to Children’s Mercy NICU 2010-2016. Inclusion criteria required infants to be preterm, admitted within three days of life, and intubated within one day post-admission. Infants must also have been intentionally extubated and subsequently reintubated prior to 36-weeks, those with multiple extubations before 36-weeks were excluded. BPD status at 36-weeks was classified per the CHND; and grouped into (None/MildBPD) and (Severe-BPD/Death) for analysis. Automated methodology for extracting periods of NIV support from EMR data was developed and validated on a random 30 infants (~5%). For each infant, timestamped data was extracted, including measures of respiratory support (FiO2) and physiologic measurements (blood gases, respiratory rate, SpO2). Bayesian logistic regression was used to estimate the magnitude and directionality of association between NIV measurements and BPD outcomes. Models were adjusted for known confounders, including demographics (gestational age, birthweight, sex, race), use of surfactant or antenatal steroids, and timing (extubation day of life, hours extubated, and time reintubated before assessment).

Results: All infants with complete data were utilized, resulting in a cohort of 124. Beyond known relationships to BPD (e.g., sex, length of time reintubated), we identified novel associations between the maximal and minimal values of blood gas measures during NIV, and BPD Outcomes. Complete results can be found in Table 1.

Conclusions: This work offers a first step to understanding how patient deterioration during NIV may be associated with longerterm outcomes. Future work includes more comprehensive characterization of NIV blood gas trajectories, and exploration into high-risk thresholds. Understanding these relationships may aid in standardization of reintubation strategies to improve patient morbidity.