ABSTRACT
Background
Bronchopulmonary dysplasia (BPD), defined as need for oxygen/respiratory support at 36 weeks gestational age (GA) is associated with increased risk of post‐prematurity respiratory disease (PRD). We hypothesize that BPD, higher pCO2, and pulmonary hypertension (PH) before NICU discharge will predict PRD.
Objectives
(1) Identify clinical factors before NICU discharge associated with PRD by 2 years of age; (2) Identify clinical factors associated with emergency room (ER) visits by 2 years of age; (3) Compare predictive performance for PRD of individual and multivariable clinical factors.
Methodology
Children born < 29 weeks GA with ≥ 1 echocardiogram before NICU discharge at two tertiary centers were included. Retrospective chart review included clinical factors at NICU discharge, ER visits, and respiratory‐related hospitalizations by 2 years. Analysis of predictors included logistic regression and ROC.
Results
We included 125 premature infants, of whom 53 (42%) had BPD, and 24 (19%) experienced PRD. All who experienced PRD had BPD. More severe BPD (OR: 96.1, CI: 12.4, 12, 383), but not hypercapnia or PH, were associated with PRD. On ROC analysis, combination of BPD severity, pCO2 and PH demonstrated 70% chance of PRD (AUC: 0.68 (95% CI: 0.55, 0.81). Presence of ≥ 2 factors had sensitivity of 50% and specificity of 97% for prediction of PRD. Children with BPD had 2.6 times as many ER visits as those without.
Conclusion
Combination of BPD severity, pCO2, and PH best predicted PRD. Identifying extremely preterm infants at high risk of developing PRD can guide counseling of families and early intervention.
Keywords: bronchopulmonary dysplasia (BPD), extremely preterm infants, post‐prematurity respiratory disease (PRD), pulmonary hypertension (PH), respiratory hospitalizations
Abbreviations
- BPD
bronchopulmonary dysplasia
- CHEO
Children's Hospital of Eastern Ontario
- ER
emergency room
- GA
gestational age
- NICU
neonatal intensive care unit
- PH
pulmonary hypertension
- PRD
post‐prematurity respiratory disease
- PROP
Prematurity and Respiratory Outcomes Program
- ROC
Receiver Operating Characteristic
- TOH
The Ottawa Hospital
1. Introduction
Bronchopulmonary dysplasia (BPD) is the most common pulmonary complication of infants who are born extremely prematurely. It is associated with significant morbidity and mortality, and its incidence is increasing due to the survival of increasingly prematurely born infants. The incidence of BPD is 44.9% in infants born less than 28 weeks gestational age (GA) and 45.2% for extremely low birth weight infants in the United Sates [1]. In Canada, among infants born at < 29 weeks' GA, the incidence of chronic lung disease (which includes BPD) was 59.2% [2]. BPD is characterized by an arrest in lung growth, leading to alveolar simplification and pulmonary vascular dysangiogenesis [3].
The respiratory health trajectory of individual infants with BPD is difficult to predict. It can be characterized by post‐prematurity respiratory disease (PRD) in the first 2 years of life [4]. PRD is a new outcome defined by the Prematurity and Respiratory Outcomes Program (PROP) and captures respiratory morbidity when the infant has at least one of the following characteristics: ≥ 2 respiratory‐related hospitalizations, home oxygen or any home respiratory support at 3 months corrected age, systemic corticosteroids, or pulmonary vasodilators after discharge, and/or death secondary to a cardiopulmonary cause [4]. Neonatal risk factors for PRD include exposure to mechanical ventilation and postnatal steroids, maternal atopy, and asthma [5]. Infants born extremely prematurely may have an abnormal trajectory of their pulmonary function, as seen by a decreased rate of rise in lung function during childhood and adolescence, abnormal peak lifetime pulmonary function, and a more rapid decline in lung function in adulthood [5]. However, prediction of PRD from clinical parameters in early in life has been variable to date [6].
A new classification system for BPD has been defined by the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network, published by Jensen et al. in 2019 [7], which groups infants based on respiratory support required at 36 weeks postmenstrual age: no BPD = no support; Grade 1 = nasal cannula oxygen at ≤ 2 L/min; Grade 2 = nasal cannula > 2 L/min or non‐invasive positive airway pressure; Grade 3 = invasive mechanical ventilation [7]. This new definition of BPD is advantageous because it more accurately predicts adverse respiratory and neurodevelopmental outcomes than previous definitions [7]. Using the Jensen classification, infants with greater severity of BPD are at a higher risk of health complications after neonatal intensive care unit (NICU) discharge and greater risk of rehospitalization up to 1‐year post‐discharge [8]. The predictive ability of the Jensen definition for respiratory morbidity at greater than 1 year of age has yet to be evaluated.
Several independent risk factors have been identified in the prediction of longer‐term outcomes of infants with BPD, some of which are identifiable just before discharge from the NICU, including pulmonary hypertension (PH) and hypercapnia (Figure 1 [9, 10, 11, 12, 13, 14]). PH associated with BPD is associated with high morbidity and mortality [13]. Approximately 25% of infants with moderate to severe BPD develop late PH, which increases mortality; 47% of BPD infants die within 2 years after diagnosis of PH [13]. Additionally, BPD‐PH is associated with higher rates of tracheostomy, increased use of supplemental oxygen, feeding problems, and frequent hospital admissions [13]. Another predictor of the prognosis of infants with BPD is pCO2. Elevated pCO2 is associated with an increased risk of adverse events, including reintubation and readmission to hospital in infants with BPD [15]. High pCO2 during non‐invasive respiratory support at 36 weeks postmenstrual age in severe BPD is also associated with rehospitalization [14]. Although there is no strong correlation between the number of ER visits and the degree of severity of BPD [16] infants with BPD are at an increased risk of post‐discharge respiratory illness, resource utilization, ER visits, and rehospitalization [17]. Infants discharged on home oxygen had more respiratory readmissions than those discharged on room air [15].
Figure 1.
This directed acyclic graph (DAG) depicts potentially caucal factors in the pathway from extreme prematurity to severe post‐prematuirty respiratory disease (PRD, color orange). This DAG was used to guide a priori selection of candidate predictors, indentifiable close to NICU discharge, which are used in our modeling (color green). [Color figure can be viewed at wileyonlinelibrary.com]
Previous studies in infants with BPD have evaluated several risk factors for post‐discharge respiratory morbidity with variable results but have not used the Jensen definition of BPD, specifically in relation to PRD, nor have they considered multiple predictors in combination. Improved identification of risk factors at the time of initial discharge from the NICU, which can predict severe respiratory morbidity in early life, would allow clinicians and families to better prepare for and mitigate potential adverse events with earlier referral and closer follow‐up. Whereas previous studies have focused on maternal and perinatal parameters identifiable at the time of birth [6] we hypothesized that clinical factors which represent the infant′s early life clinical course and are measured after the acute phase of care, before first hospital discharge, would be predictive of respiratory morbidity. Therefore, pCO2 levels and BPD‐PH in addition to Jensen's BPD severity were the focus of this study as they are factors that are measurable at the time of NICU discharge.
This study's primary objective was to identify clinical factors before initial NICU discharge associated with severe PRD up to 2 years corrected age in children. The secondary objective was to identify clinical factors associated with the number of emergency room (ER) visits for respiratory symptoms up to 2 years of life. The exploratory objective was to compare the predictive performance of individual predictors of severe PRD, including BPD diagnosis, the number of ER visits, the pCO2 level before discharge from initial NICU stay, and the presence of PH, as well as to evaluate the predictive performance of the combination of multiple risk factors.
2. Methods
2.1. Study Design and Setting
This study included children from a previous cohort study, born at < 29 weeks GA between June 1, 2012, and April 18, 2016, who were admitted to either the Children's Hospital of Eastern Ontario (CHEO) NICU or The Ottawa Hospital (TOH) NICU, underwent ≥ 1 echocardiogram before first hospital discharge and survived to first hospital discharge [18, 19]. This study built upon this study, collecting follow‐up data on these infants until 2 years of life. Patients who died before initial NICU discharge and those transferred to a hospital outside of Ottawa, Ontario, before the first hospital discharge were excluded from this study. For this study, the cohort was previously conceived and therefore the study sample of 125 was predetermined. A priori calculations determined that given the anticipated event rate of PRD (35% from the PROP study) [6] to four single‐degree of freedom predictor varaibles could be included, using the rule of thumb of 10 events per model degree of freedom.
Patient charts were reviewed to extract data on demographics, laboratory investigations, including blood gases and echocardiograms, and respiratory morbidity outcomes. Data was stored in a secure REDCap database. Variables explored as candidate predictors of PRD included diagnosis of BPD according to the Jensen definition [7], number of ER visits after discharge in the first 2 years of life, elevated CO2, and presence of PH. We defined PH by the presence of any of the following criteria: (a) elevated right ventricular pressures as estimated by Doppler studies of tricuspid regurgitation jet‐based measurement of right ventricular systolic pressure > half arterial systolic pressure or (b) flattening of the interventricular septum in end‐systole and (c) any evidence of right‐to‐left shunting [20]. pCO2was considered elevated if levels were > 50 mmHg. All infants in the study underwent echocardiogram in the first week of life. Those with elevated pulmonary pressures subsequently underwent serial echocardiograms every 2 weeks until resolution or hospital discharge. Since both early (at 1 week of life) and late PH (after 36 weeks) have been shown to be predictors of moderate to severe BPD [21, 22], we hypothesized that both could be important predictors of PRD and they were considered separately in modeling. A priori hypotheses and the number of predictors evaluated were restricted to these variables to ensure that there would be sufficient power with our sample size to evaluate the factors of greatest interest. This study was approved by the Research Ethics Board (REB) at the Children's CHEO (REB Protocol No: CHEOREB#23/51X). Informed consent was waived, as the REB′s criteria for a waiver were fully met.
2.2. Study Outcomes
The primary outcome of interest of this study was the presence/absence of PRD within the first 2 years of life using the PROP definition of PRD, which states that PRD is present if the child has any of the following: ≥ 2 respiratory‐related hospitalizations, home oxygen or any home respiratory support at 3 months corrected age, systemic corticosteroids, or pulmonary vasodilators after discharge, and/or death secondary to a cardiopulmonary cause [4]. Additional outcomes of interest included the number of ER visits and hospitalizations for respiratory symptoms up to 2 years. Information on specific respiratory symptoms and diagnoses at presentation to the ER or hospital was also collected.
2.3. Statistical Analysis
The association between the primary outcome (PRD) and the clinical factors, including the presence/absence of BPD, pCO2 level (mmHg), and presence of PH pre‐discharge, was assessed using univariate and multivariable logistic regression with Firth correction. The association between the Jensen grade severity of BPD and PRD outcome was assessed using a univariate logistic regression. When the study was designed, we anticipated analyzing severity of BPD using “no BPD” (Jensen grade 0) as the reference category. We found, however that none of the infants in our cohort without BPD developed PRD and therefore the planned analysis could not be carried out. We therefore used Jensen grade 0−1 as the reference category. The secondary analysis assessed the association of the same clinical factors with ER visits in the first 2 years of life, using a negative binomial regression model. Lastly, a Receiver Operating Curve (ROC) analysis was conducted to explore the ability of multiple risk factors, including PH pre‐discharge, a Jensen Score ≥ 2, and a pCO2 level > 50 mmHg, to distinguish PRD cases. Odds Ratios (for logistic regressions) and Rate Ratios (for negative binomial) with associated 95% confidence intervals (CI) were reported. A two‐sided p < 0.05 was considered statistically significant. All analyses were conducted using R statistical computing software (Version 4.3.1) [23].
3. Results
A total of 125 infants were included in the study (Figure 2). Table 1 describes the baseline characteristics of the study population. Fifty‐three of the infants in the study (42%) had BPD, and 72 (58%) had no BPD. Those with BPD were further classified based on the Jensen definition: 30 babies (57%) had Grade 1 BPD, 18 (34%) had Grade 2 BPD, and 5 (9%) had Grade 3 BPD. Children with BPD had lower GA and birth weight, longer duration of respiratory support, and higher prevalence of retinopathy of prematurity (Table 1). Additionally, at initial discharge from the NICU, those with BPD tended to be older. There was no difference in the incidence of PH between those with BPD and those without BPD.
Figure 2.
CONSORT flowchart outlining the size of intial patient population, how many were excluded and the final sample size.
Table 1.
Characteristics of the population of the study. N = 125.
| Variable | N | Overall N = 125a | No BPD N = 72 | BPD N = 53 | p valueb | Adjusted p valuec |
|---|---|---|---|---|---|---|
| Gestational age (weeks) | 125 | < 0.001 | < 0.001 | |||
| Median (Q1, Q3) | 27.1 (25.9, 28.1) | 27.7 (26.5, 28.3) | 26.0 (25.0, 27.3) | |||
| Mean (SD) | 26.8 (1.5) | 27.3 (1.3) | 26.2 (1.5) | |||
| Corrected gestational age (weeks) at discharge | 124 | < 0.001 | < 0.001 | |||
| Median (Q1, Q3) | 37.6 (34.6, 40.4) | 35.4 (33.3, 37.7) | 40.4 (39.1, 42.0) | |||
| Mean (SD) | 37.7 (4.5) | 35.4 (3.1) | 40.7 (4.2) | |||
| Missing | 1 | 1 | 0 | |||
| Sex | 125 | 0.37 | > 0.99 | |||
| Male | 61 (48.8%) | 38 (52.8%) | 23 (43.4%) | |||
| Female | 64 (51.2%) | 34 (47.2%) | 30 (56.6%) | |||
| Birth weight (g) | 125 | < 0.001 | < 0.001 | |||
| Median (Q1, Q3) | 970.0 (770.0, 1175.0) | 1120.0 (895.0, 1275.0) | 830.0 (689.0, 1000.0) | |||
| Mean (SD) | 979.7 (267.2) | 1,079.3 (253.9) | 844.4 (223.3) | |||
| Pulmonary hypertension at any time | 125 | 75 (60.0%) | 38 (52.8%) | 37 (69.8%) | 0.066 | 0.53 |
| Pulmonary Hypertension predischarge | 125 | 32 (25.6%) | 17 (23.6%) | 15 (28.3%) | 0.68 | > 0.99 |
| Number of days on oxygen | 73 | < 0.001 | 0.002 | |||
| Median (Q1, Q3) | 21.3 (10.1, 39.0) | 11.0 (3.1, 22.6) | 29.0 (12.3, 43.0) | |||
| Mean (SD) | 24.1 (17.6) | 14.2 (12.1) | 29.3 (17.9) | |||
| Missing | 52 | 47 | 5 | |||
| Number of days on ventilation | 123 | < 0.001 | < 0.001 | |||
| Median (Q1, Q3) | 37.9 (11.1, 59.6) | 17.6 (6.8, 38.0) | 64.2 (42.9, 78.8) | |||
| Mean (SD) | 40.7 (30.7) | 23.3 (19.0) | 63.5 (28.3) | |||
| Missing | 2 | 2 | 0 | |||
| Patent ductus arteriosus | 125 | 98 (78.4%) | 53 (73.6%) | 45 (84.9%) | 0.19 | > 0.99 |
| Necrotizing enterocolitis | 125 | 10 (8.0%) | 3 (4.2%) | 7 (13.2%) | 0.095 | 0.66 |
| Intraventricular hemorrhage | 125 | 52 (41.6%) | 29 (40.3%) | 23 (43.4%) | 0.85 | > 0.99 |
| Retinopathy of prematurity | 125 | 35 (28.0%) | 13 (18.1%) | 22 (41.5%) | 0.005 | 0.049 |
| Congenital diseases | 125 | 22 (17.6%) | 10 (13.9%) | 12 (22.6%) | 0.24 | > 0.99 |
| Sepsis | 125 | 28 (22.4%) | 10 (13.9%) | 18 (34.0%) | 0.010 | 0.086 |
| Genetic syndrome | 125 | 8 (6.4%) | 3 (4.2%) | 5 (9.4%) | 0.28 | > 0.99 |
Note: Wilcoxon rank sum test for all continuous variables.
n (%).
Fisher's test for all categorical variables.
Adjust p values using Holm's method.
Twenty‐Four (19%) experienced a PRD event, whereas 101 (81%) did not. All the infants who experienced PRD had BPD (Table 2).
Table 2.
Associations between candidate predictors and PRD.
| Variable | PRD, N = 24 | No PRD, N = 101 |
|---|---|---|
| BPD | ||
| No BPD | 0 (0.0%) | 72 (71.3%) |
| BPD | 24 (100.0%) | 29 (28.7%) |
| PCO2 level (mmHg) | ||
| Median (IQR) | 51.0 (46.0, 61.0) | 47.0 (41.0, 52.3) |
| Mean (SD) | 52.5 (9.1) | 46.7 (7.6) |
| Missing | 0 | 1 |
| Pulmonary hypertension at any time point | 18 (75.0%) | 57 (56.4%) |
| Pulmonary hypertension pre‐discharge | 9 (37.5%) | 23 (22.8%) |
3.1. Oxygen Requiement at Discharge
Of the 125 partcipants in the study, 22 (17.6%) were discharged on home oxygen. Of those who had PRD (n = 24), 20 (80.3%) infants were discharged on home oxygen. Of those with no PRD (101) 2 (2.0%) were discharged on home oxygen. The median (IQR) oxygen prescription at discharge was 0.13 L/min (0.05, 0.13) for those with PRD and 0.03 (0.03, 0.03) for those with no PRD.
3.2. BPD Severity
BPD severity grade was stratified by PRD outcome in the 53 infants with BPD. All the infants with PRD had BPD. Of the 24 infants who experienced the PRD outcome, 14 (58.3%) had Grade 1 BPD, 7 (29.2%) had Grade 2 BPD, and 3 (12.5%) of them had Grade 3 BPD. Of the 29 infants with BPD who did not experience the PRD outcome, 16 (55.2%) had Grade 1 severity of BPD, 11 (37.9%) had Grade 2 severity of BPD, and 2 (6.9%) had Grade 3 severity of BPD. Table 3 shows the results of a univariate logistic regression of PRD with Jensen definition for infants with BPD, with no differences detected between the severity grade of BPD and PRD outcome.
Table 3.
Results of univariate logistic regression of PRD with Jensen score categories for infants with BPD only.
| Variable | N | OR | 95% CI | p value |
|---|---|---|---|---|
| Jensen score | 0.68 | |||
| Grade 1 | 30 | — | — | |
| Grade 2 | 18 | 0.73 | 0.21, 2.37 | |
| Grade 3 | 5 | 1.71 | 0.25, 14.5 |
Abbreviations: CI, confidence interval; OR, odds ratio.
The proportion of infants who had ER visits within the first 2 years of life was compared between those with BPD and those without BPD. Figure 3 demonstrates that infants with BPD tended to have more ER visits compared to those without BPD. A negative binomial regression of ER visits with BPD revealed that infants with BPD have 2.6 times more ER visits over the first 2 years of life than those without BPD (CI 1.47−4.67; p = 0.001).
Figure 3.
Count of total ER visits for infants with and without BPD over the first two years of life. N = 125. [Color figure can be viewed at wileyonlinelibrary.com]
3.3. Pco2 Level
The pCO2 levels before discharge from the NICU were widely distributed, ranging from 32.0 to 68.0 mmHg. At hospital discharge, 79/125 (64%) of infants had pCO2 levels ≤ 50 mmHg, and 45/125 (36%) had pCO2 levels > 50 mmHg. One infant was missing pCO2 data. Figure 4 demonstrates the association between pCO2 level before discharge and the PRD outcome. The mean pCO2 level at discharge tended to be higher amongst those who developed PRD compared to those who did not have PRD (mean, SD 52.5 ± 9.1 mmHg vs. 46.7 ± 7.6 mmHg; p = 0.2). Of those who had pCO2 levels > 50 mmHg, in 37 (84.09%) it was drawn from a capillary blood sample, three infants (6.82%) had venous blood samples, and four infants (9.09%) had arterial blood drawn. No significant associations were found between pCO2 level and PRD median difference of 4.0, 95% CI of 0.98−1.12, and p value of 0.2).
Figure 4.
pCO2 level before discharge and PRD outcome (median difference of 4.0, 95% CI of 0.98−1.12, and p value of 0.2). [Color figure can be viewed at wileyonlinelibrary.com]
3.4. PH
PH at anytime of the infants' NICU stay was present in 75/125 (60%) in this cohort and was more common in those with BPD (Table 1). Forty‐eight infants out of the full cohort (38%) had early/early persistent PH; 21 (44%) of those infants persisted to have late PH. Eleven babies without early/early persistent PH were found to have PH at 36 weeks, making the total number of infants with late PH 32. In univariable and multivariable logistic regression modeling with Firth correction, PH was not, however, found to be significantly associated with PRD outcome (Table 4).
Table 4.
Results of univariable and multivariable logistic regression of PRD outcome (univariate BPD and full multivariable model was fitted using logistic regression with Firth Correction).
| Univariable | Multivariable | ||||
|---|---|---|---|---|---|
| Variable | OR | 95% CI | OR | 95% CI | p value |
| BPD | |||||
| No BPD | — | — | — | — | |
| BPD | 120 | 15.8, 15,467 | 96.1 | 12.4, 12,383 | < 0.001 |
| PCO2 level (mmHg) | 1.09 | 1.03, 1.16 | 1.05 | 0.98, 1.12 | 0.2 |
| Pulmonary hypertension pre‐discharge | |||||
| No | — | — | — | — | |
| Yes | 2.03 | 0.77, 5.21 | 2.28 | 0.69, 8.03 | 0.2 |
Abbreviations: CI, confidence interval; OR, odds ratio.
3.5. Multiple Predictors
Finally, the presence of multiple predictors was examined simultaneously in the context of PRD, including PH pre‐discharge, a Jensen Score ≥ 2, and a pCO2 level > 50 mmHg. Table 5 and Figure 5 show the proportion of the study population with none, one, or two or more factors and the respective likelihood of PRD outcome. These findings demonstrate that a greater number of predictors is associated with a higher risk of having PRD. A ROC analysis of PRD and the number of predictors was performed, as seen in Figure 6. The area under the curve is 0.680, and the point highlighted on the figure represents having 2 or more of the factors corresponding to a specificity of 87% and a sensitivity of 50%.
Table 5.
Associations between the presence of none, one, two, or all the following (pCO2 > 50, pulmonary hypertension pre‐discharge, Jensen score ≥ 2) and PRD.
| Variable | N | PRD, N = 24 | 95% CI | No PRD, N = 101 | 95% CI |
|---|---|---|---|---|---|
| None of these factors | 125 | 7 (29.17%) | 15%, 49% | 47 (46.53%) | 37%, 56% |
| One of these factors present | 125 | 5 (20.83%) | 9.2%, 40% | 41 (40.59%) | 32%, 50% |
| Two factors present | 125 | 9 (37.50%) | 21%, 57% | 12 (11.88%) | 6.9%, 20% |
| Three factors present | 125 | 3 (12.50%) | 4.3%, 31% | 1 (0.99%) | 0.17%, 5.4% |
| Two factors or more | 125 | 12 (50.00%) | 31%, 69% | 13 (12.87%) | 7.7%, 21% |
Abbreviation: CI, confidence interval.
Figure 5.
Proportions with 95% CI for having PRD or not for infants with none, one, two, or more of the following factors: pulmonary hypertension pre‐discharge, Jensen score ≥ 2, and pCO2 > 50 mmHg. [Color figure can be viewed at wileyonlinelibrary.com]
Figure 6.
ROC curve of PRD and the number of factors (pCO2 > 50, Jensen score ≥ 2, and PH pre‐dishcarge). The point highlighted on the figure represents having 2 or more of the factors corresponds to a specificity of 87% and a sensitivity of 50%.
4. Discussion
This study found that PRD occurred exclusively in infants with BPD and that infants with BPD were 2.6 times more likely to have respiratory‐related ER visits in the first 2 years of life, compared to those without BPD. Individual clinical factors including severity grade for BPD, pCO2 and PH did not independently predict PRD, although mean pCO2 level before hospital discharge tended to be higher in those who developed PRD than those who did not. Examination of multiple clinical factors in combination to predict PRD is a novel feature of this study. We found that as the number of these clinical risk factors increased, so did the infant's risk of PRD.
We found that presence of BPD is a strong predictor for the development of PRD, as all the infants who developed PRD also had BPD, although not all infants with BPD developed PRD. However, severity grade of BPD did not predict PRD outcome in our study. Although the classification of BPD defined by Jensen et al. is advantageous in informing respiratory care practices and can be used to predict early childhood morbidity, it has some limitations that make it not discriminatory in predicting PRD over 2 years. This may be related to the definition's dependence on clinical parameters, including the need for oxygen and respiratory support, rather than on the pathophysiology of the disease. As a result, some patients may be misclassified as having BPD when they have different pathology, such as congenital heart disease or upper airway anomalies [24].
Some studies show that those with BPD are at an increased risk for ER visits [16, 17]. Our findings are consistent with what is present in the literature, as children in our study with BPD had 2.6 times more ER visits compared to those without BPD.
Elevated levels of pCO2 have been associated with an increased risk of adverse events, such as reintubation and readmission to the hospital [15]. Our analysis revealed that the mean pCO2 level at discharge tended to be higher among infants who developed PRD compared to those who did not develop PRD. Previous studies have examined pCO2 levels before discharge as a predictor for rehospitalization and a need for home oxygen [17] and revealed that children with BPD and elevated pCO2 levels had an increased risk of reintubation, and readmission to hospital. To our knowledge, this is the first study to look at the pCO2 level before discharge in relation to the PRD outcome in the first 2 years of life.
Our analysis did not reveal any significant associations between PH and PRD. This may be because PH is relatively common in early life in extremely low birth weight infants and in infants with severe BPD. Our prevalence of PH at any time during NICU stay of 60% is consistent with literature indicating that prevalence of early PH in extremely preterm infants, can be up to 55% [7, 22, 25]. Since our echocardiograms were conducted within the first week of life, it is not surprising that our early PH prevalence was higher than that reported in a similar cohort within the first 2 weeks of life [26]. Similarly, the prevalence of late PH (beyond 36 weeks gestation) in our cohort (26%) is consistent with that reported in the literature (18% in moderate BPD and 41% in severe BPD [22, 26]. However, it often resolves in infants that survive more than 6 months, therefore no longer impacting the infant's health outcomes [27]. Long‐term impacts on health outcomes in infants with BPD‐PH are lacking in the literature [28, 29], and this will require further exploration in larger, prospective studies. Further, although there were no differences in PH prevalence at any time point or predischarge between babies who developed BPD and those who did not, this may be because the majority of our cohort consisted of infants without or with milder severity of BPD. Nonetheless, our prevalence of early and late PH in those born extremely preterm is consistent with that in a previous cohort of infants born extremely preterm [30].
Notably, we found that simultaneous consideration of multiple clinical factors increased the likelihood of predicting PRD. The presence of PH pre‐discharge, Jensen score ≥ 2, and pCO2 > 50 mmHg in combination best predicted PRD in our study. The ROC curve analysis revealed approximately 70% chance that the model will be able to distinguish between a baby who will develop PRD and those who will not. This makes it a potentially useful tool for advising families about future risks of respiratory problems and targeting post‐discharge support to those at greatest risk of morbidity and return hospital visits.
Our study does have several limitations. One pertains to the clinical definition of BPD, which has been evolving and changing over the years, so what we define as BPD in our study may differ from what is classified as BPD in other literature. Additionally, the inability of BPD severity to predict PRD may be due to the small sample size, as only 5 patients in the study had Jensen's Grade 3 severity of BPD. Furthermore, as the definition is based on the need for supplemental oxygen and respiratory support, it may be capturing different pathologies and, therefore, misclassifying some infants as having BPD. Another limitation is that ours was a retrospective study. As a result, there was some missing data in the charts, especially as some infants were transferred to hospitals in their hometowns and were lost to follow‐up. Neonatal data was not available on infants who were not included in this cohort. Additionally, the study sample may not be representative of the true cohort, as sicker patients were followed at our tertiary care center and included in this study. In contrast, children who were more well may have received care closer to home and may not have had follow‐ups at our center after their first hospital discharge. The variablility in the timing of the pre‐discharge pCO2 is another limitation. We used the last pCO2 before discharge, but as this was a retrospective study, the timing of the last capillary blood gas was not standardized. Once capillary blood gas normalized (< 50 mmHg) it was not repeated. Finally, the timing of echocardiogram was not fully standardized in this retrospective study. Nonetheless, all infants underwent echocardiogram evaluations within the first week of life and those with elevated pulmonary pressures underwent longitudinal follow up studies until PH resolved.
In conclusion, this study explored several risk factors in predicting longer‐term post‐discharge respiratory morbidity, specifically using newer classifications of BPD and PRD. This is also the first study to examine the joint effects of multiple risk factors to predict PRD. We can conclude that BPD is a strong risk factor for the development of PRD, although the severity grade of BPD was not associated with increased PRD risk. Furthermore, the key finding of this study was that a combination of risk factors (pCO2 > 50, BPD grade ≥ 2, and PH on predischarge) increased the risk of a child developing BPD and, therefore, can be used as a predictive model for PRD. Identification of extremely preterm infants who are at high risk of developing PRD can help guide the counseling of families and follow‐up of the highest‐risk infants. Larger, prospective studies are needed to further understand risk factors for long‐term respiratory morbidity in this high‐risk population. An additional potential future direction would be to compare babies with BPD who develop PRD to those with BPD who did not, to learn about susceptibility/resilience factors in babies with BPD who develop PRD.
Author Contributions
Sophie Holcik: writing − review and editing, writing − original draft, data curation, investigation, project administration. lamia hawayi: methodology, writing − review anandd editing, formal analysis. Naomi Dussah: data curation, writing − review and editing, project administration. Nick Barrowman: writing − review and editing, methodology, formal analysis, supervision. Nadya Ben Fadel: investigation, writing − review and editing, supervision. Bernard Thébaud: investigation, writing − review and editing, supervision. Sherri Lynne Katz: conceptualization, investigation, writing − original draft, writing − review and editing, supervision.
Conflicts of Interest
The authors declare no conflicts of interest.
Data Availability Statement
Data sharing not applicable to this article as no data sets were generated or analyzed during the current study.
References
- 1. Lee S. M., Sie L., Liu J., Profit J., and Lee H. C., “Evaluation of Trends in Bronchopulmonary Dysplasia and Respiratory Support Practice for Very Low Birth Weight Infants: A Population‐Based Cohort Study,” Journal of Pediatrics 243 (2022): 47–52.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. The Canadian Neonatal Network . The Canadian Neonatal Network TM 2023 Annual Report Le Réseau Néonatal Canadien TM 2023 Rapport Annuel i. 2023.
- 3. Thébaud B., Goss K. N., Laughon M., et al., “Bronchopulmonary Dysplasia,” Nature Reviews Disease Primers 5, no. 1 (2019): 78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Pryhuber G. S., Maitre N. L., Ballard R. A., et al., “Prematurity and Respiratory Outcomes Program (PROP): Study Protocol of a Prospective Multicenter Study of Respiratory Outcomes of Preterm Infants in the United States,” BMC Pediatrics 15, no. 1 (April 2015): 37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Levin J. C., Sheils C. A., Gaffin J. M., Hersh C. P., Rhein L. M., and Hayden L. P., “Lung Function Trajectories in Children With Post‐Prematurity Respiratory Disease: Identifying Risk Factors for Abnormal Growth,” Respiratory Research 22, no. 1 (May 2021): 143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Keller R. L., Feng R., DeMauro S. B., et al., “Bronchopulmonary Dysplasia and Perinatal Characteristics Predict 1‐Year Respiratory Outcomes in Newborns Born Atat Extremely Low Gestational Age: A Prospective Cohort Study,” Journal of Pediatrics 187 (August 2017): 89–97.e3.e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Jensen E. A., Dysart K., Gantz M. G., et al., “The Diagnosis of Bronchopulmonary Dysplasia in Very Preterm Infants. An Evidence‐Based Approach,” American Journal of Respiratory and Critical Care Medicine 200, no. 6 (2019): 751–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Lagatta J. M., Zhang L., Yan K., et al., “Prospective Risk Stratification Identifies Healthcare Utilization Associated With Home Oxygen Therapy for Infants With Bronchopulmonary Dysplasia,” The Journal of Pediatrics 251 (December 2022): 105–112.e1. [DOI] [PubMed] [Google Scholar]
- 9. Jain D., Feldman A., and Sangam S., “Predicting Long‐Term Respiratory Outcomes In Premature Infants: Is it Time to Move Beyond Bronchopulmonary Dysplasia?,” Children (Basel) 7, no. 12 (2020): 283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Greenough A., Limb E., Marston L., Marlow N., Calvert S., Peacock J., “Risk Factors for Respiratory Morbidity in Infancy After Very Premature Birth,” Archives of Disease in Childhood. Fetal and Neonatal Edition 90, no. 4 (July 2005): F320‐3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Sun S., Zivanovic S., Earnest A., Roehr C. C., and Tan K., “Respiratory Management and Bronchopulmonary Dysplasia in Extremely Preterm Infants: A Comparison of Practice Between Centres in Oxford and Melbourne,” Journal of Perinatology 42 (January 2022): 53–57. [DOI] [PubMed] [Google Scholar]
- 12. Weinstock J., Xuchen X., Arroyo M., et al., “The Next Frontier of Prematurity: Predicting Respiratory Morbidity During the First Two Years of Life in Extremely Premature Babies,” Cureus 14, no. 3 (March 2022): e23505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hansmann G., Sallmon H., Roehr C. C., Kourembanas S., and Austin E. D., “Koestenberger M. Pulmonary Hypertension in Bronchopulmonary Dysplasia.” Pediatric Research (Springer Nature, 2021). 89, 446–455. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Shin S. H., Shin J., Kim E. K., and Kim H. S., “Capillary Partial Pressure of Carbon Dioxide for Predicting Rehospitalization in Preterm Infants Under Noninvasive Respiratory Support With Severe Bronchopulmonary Dysplasia,” Pediatric Pulmonology 56, no. 12 (2021): 3863–3869. [DOI] [PubMed] [Google Scholar]
- 15. Kovesi T., Abdurahman A., and Blayney M., “Elevated Carbon Dioxide Tension as a Predictor of Subsequent Adverse Events in Infants With Bronchopulmonary Dysplasia,” Lung 184, no. 1 (2006): 7–13. [DOI] [PubMed] [Google Scholar]
- 16. Kurihara C., Zhang L., and Mikhael M., “Newer Bronchopulmonary Dysplasia Definitions and Prediction of Health Economics Impacts in Very Preterm Infants,” Pediatric Pulmonology 56, no. 2 (2021. Feb): 409–417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Vohr B., McGowan E., Keszler L., O'Donnell M., Hawes K., and Tucker R., “Effects of a Transition Home Program on Preterm Infant Emergency Room Visits Within 90 Days of Discharge,” Journal of Perinatology 38, no. 2 (2017): 185–190. [DOI] [PubMed] [Google Scholar]
- 18. Hanna Y., Laliberté C., Fadel N., et al., Effect of Oxygen Saturation Targets on the Incidence of Bronchopulmonary Dysplasia and Duration of Respiratory Supports in Extremely Preterm Infants. [DOI] [PMC free article] [PubMed]
- 19. Laliberté C., Hanna Y., Ben Fadel N., et al., “Target Oxygen Saturation and Development of Pulmonary Hypertension and Increased Pulmonary Vascular Resistance in Preterm Infants,” Pediatric Pulmonology 54, no. 1 (January 2019): 73–81. [DOI] [PubMed] [Google Scholar]
- 20. Giesinger R. E., More K., Odame J., Jain A., Jankov R. P., and McNamara P. J., “Controversies in the Identification and Management of Acute Pulmonary Hypertension in Preterm Neonates.” Pediatric Research (Nature Publishing Group, 2017). 82, 901–914. [DOI] [PubMed] [Google Scholar]
- 21. Mourani P. M., Sontag M. K., Younoszai A., et al., “Early Pulmonary Vascular Disease in Preterm Infants at Risk for Bronchopulmonary Dysplasia,” American Journal of Respiratory and Critical Care Medicine 191, no. 1 (January 2015): 87–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Mascarenhas D., Al‐Balushi M., Al‐Sabahi A., Weisz D. E., Jain A., and Jasani B., “Pulmonary Hypertension in Preterm Neonates With Bronchopulmonary Dysplasia: A Meta‐Analysis,” ADC Fetal & Neonatal 110, no. 4 (June 2025): 344–352. [DOI] [PubMed] [Google Scholar]
- 23. Core Team R., R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2022). [Google Scholar]
- 24. Collaco J. M. and McGrath‐Morrow S. A., “Respiratory Phenotypes for Preterm Infants, Children, and Adults: Bronchopulmonary Dysplasia and More,” Annals of the American Thoracic Society 15 (2018): 530–538. [DOI] [PubMed] [Google Scholar]
- 25. Mani S., Mirza H., Ziegler J., and Chandrasekharan P., “Early Pulmonary Hypertension in Preterm Infants,” Clinics in Perinatology 51 (2024): 171–193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Kim Y. J., Shin S. H., Park H. W., Kim E. K., and Kim H. S., “Risk Factors of Early Pulmonary Hypertension and its Clinical Outcomes in Preterm Infants: A Systematic Review and Meta‐Analysis,” Scientific Reports 12, no. 1 (December 2022): 14186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Chan S., Brugha R., Quyam S., and Moledina S., “Diagnosis and Management of Pulmonary Hypertension in Infants With Bronchopulmonary Dysplasia: A Guide for Paediatric Respiratory Specialists,” Breathe. European Respiratory Society 18, no. 4 (2022): 220209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Kim C., Kim S., Kim H., et al, “Long‐Term Impact of Late Pulmonary Hypertension Requiring Medication in Extremely Preterm Infants With Severe Bronchopulmonary Dysplasia,” Scientific Reports 14, no. 1 (April 2024): 8705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Berkelhamer S. K., Mestan K. K., and Steinhorn R. H., “Pulmonary Hypertension in Bronchopulmonary Dysplasia,” Seminars in Perinatology 37 (2013): 124–131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Arjaans S., Zwart E. A. H., Roofthooft M., Kooi E. M. W., Bos A. F., and Berger R. M. F., “Pulmonary Hypertension in Extremely Preterm Infants: A Call to Standardize Echocardiographic Screening and Follow‐Up Policy,” European Journal of Pediatrics 180 (2021): 1855–1865, 10.1007/s00431-021-03931-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data sharing not applicable to this article as no data sets were generated or analyzed during the current study.