Bronchopulmonary dysplasia (BPD) is a chronic pulmonary disease that affects approximately 50% of preterm neonates delivered before 28 weeks of gestation and 30% of those delivered before 32 weeks.1,2 The mortality is higher in infants that develop BPD,3 and those that survive are at increased risk of pulmonary and cardiovascular disease and, above all, neurodevelopmental sequelae, all of which are associated with a poorer quality of life and an increased use of health care resources.4–6

The pathophysiology of BPD is multifactorial and has yet to be fully elucidated.7 At its core is preterm birth,8 on which various prenatal and postnatal factors act to modulate the baseline probability of developing the disease.9 Although multiple therapies have been investigated to prevent or treat BPD, a safe and effective treatment is not yet available, and evidence is mounting that there is a window of opportunity for treatment that far predates the clinical diagnosis of BPD.10 Higgins et al.9 proposed that strategies aimed at preventing BPD should be implemented prenatally or in the first days post birth, mainly around the first week of life. Such interventions have been found to contribute to reducing neonatal mortality and some complications associated with prematurity.11,12 However, they have not resulted in a significant reduction in the incidence of BPD. The increased survival of extremely preterm infants, in whom BPD is more prevalent, could partly explain these results, but they also reflect our inability to prevent the development of BPD.13,14

Thus, identifying neonates at higher risk of BPD is one of the present challenges for neonatology and has been subject to multiple studies15 with 2 main aims. The first, at the clinical level, is to enable the identification of patients in whom the benefits of currently available treatments outweigh their risks.16 The second, at the level of research, is to allow the adequate selection of candidates for new treatments in the context of clinical trials.

The aim of our study was to establish the incidence, clinical characteristics and risk factors of cases of BPD and moderate-to-severe BPD in a contemporary sample of very preterm infants, analysing prenatal variables and neonatal outcomes in the first weeks post birth.

In the period under study, a total of 241 neonates born preterm at or before 32 weeks of gestation with a birthweight of 1500 g or less. We excluded those with major congenital malformations (n = 16) and those deceased before 28 days post birth, in who it was not possible to diagnose BPD (n = 23, of who 8 died from respiratory causes). The final sample included 202 very preterm infants (106 female, 52.5%) (Fig. 1). The mean gestational age was 29.5 ± 2.1 weeks and the mean birthweight 1142.1 ± 255.5 g. Table 1 presents the main perinatal characteristics of the sample and the most relevant neonatal diagnosis and outcomes.

Figure 1.

Flowchart of the study sample. BPD, bronchopulmonary dysplasia; NICU, neonatal intensive care unit.

(0.13MB).

A total of 58 patients (28.7%) received a diagnosis of BPD, of who 21 (10.4% of the total) were classified as cases of moderate-severe BPD; Fig. 2 presents the distribution of cases by weeks of gestational age. Forty neonates (19.8%) required intubation in the delivery room, and 101 (50%) required surfactant during the stay. Seventy-eight (38.6%) received IMV during the stay, and the rest of the patients (n = 124) only required non-invasive respiratory support. None of the patients died after 28 days post birth.

Figure 2.

Bar graph of the distribution of patients by weeks of gestational age at birth and number of patients with BPD and moderate-severe BPD in each age group. BPD, bronchopulmonary dysplasia.

(0.13MB).

Table 2 summarises the results of the univariate analysis for BPD. The most relevant independent risk factors for BDP identified in the logistic regression analysis were gestational age (odds ratio [OR], 0.44; 95% CI = 0.30−0.65; P < 0.001), the need of IMV in the first day post birth (OR, 8.13; 95% CI = 2.41–27.42; P = 0.001), nosocomial sepsis (OR, 9.51; 95% CI = 2.99–30.28; P < 0.001) and the FiO2 on day 14 post birth (OR, 1.39; 95% CI = 1.16–1.66; P < 0.001) (Fig. 3). The AUC for the BPD model was 0.967 (95% CI, 0.944–0.990).

Figure 3.

Forest plot of the odd ratios obtained in the logistic regression models of bronchopulmonary dysplasia and moderate-to-severe bronchopulmonary dysplasia with the corresponding 95% confidence intervals. BPD, bronchopulmonary dysplasia; IMV, invasive mechanical ventilation.

(0.12MB).

Table 3 summarises the results of the univariate analysis for moderate-to-severe BPD. The most relevant independent risk factors for moderate-severe BDP identified in the logistic regression analysis were the need of IMV on days 1 (OR, 5.39; 95% CI = 1.54–18.89; P = 0.008) and 3 of life (OR, 9.99; 95% CI = 2.47–40.44; P = 0.001) and diagnosis of nosocomial sepsis (OR, 9.87; 95% CI = 2.58–37.80; P = 0.001) (Fig. 3). The AUC for the moderate-severe BPD model was 0.891 (95% CI, 0.812–0.971).

The AUCs for the ROC curves of the previously published models selected for comparison were smaller compared to the AUCs obtained in our models. In the case of Gursoy et al.,27 the model for prediction of BPD had an AUC of 0.869 (95% CI, 0.815−0.922) and the model for moderate-severe BPD an AUC of 0.803 (95% CI, 0.711−0.895). In the case of Hunt et al.,26 the model for prediction of BPD had an AUC of 0.600 (95% CI, 0.508−0.692) and the model for moderate-severe BPD an AUC of 0.624 (95% CI, 0.480−0.767).

A better understanding of the risk factors for BPD is a key step toward the prevention and adequate management of this disease. Our study, conducted in a consecutive sample of infants delivered at or before 32 weeks of gestation with birthweights of 1500 g or less, most of who were managed with non-invasive support, and with an incidence of BPD similar to the incidence described in the previous literature, corroborated that lesser GA, the need of IMV in the first days post birth and nosocomial infection are the main early risk factors for BPD. The predictive ability of our models improved compared to past models.26,27

In our patients, the incidence of moderate-severe BPD was 10.4%, lower compared to the national mean (15.8%) based on data from the SEN1500 study published by the iNeo network,11 which reported a significantly higher global incidence for participating countries (25.5%). These improved outcomes were also reflected in the survival free of moderate-severe BPD in infants delivered at or before 28 weeks, which was of 79.2% in our sample, greater than the national mean (61.4%), but similar to survival in hospitals in our region that have reported more favourable outcomes (72.5%).28 A factor that may play a role in these results is the lower frequency of extremely preterm infants in our case series.

It is widely accepted that gestational age is the best isolated predictor of BPD,7 although in most instances this is not a modifiable factor. At birth, the respiratory prognosis is often influenced by the cause of preterm birth and by the resulting interruption of foetal lung maturation.7,9 Thus, gestational birth is a very important factor in the most premature infants, who are born at a very early stage of lung maturation.7 It is highly probable that these infants will develop BPD independently of how they are managed after birth.9 Notwithstanding, there are multiple postnatal factors that can play a role in the severity of BPD and therefore influence its outcomes, especially in less preterm infants. The activation of the inflammatory cascade and oxidative stress are 2 of the key pathophysiological pathways leading to the development of BPD. In association with this physiological basis, factors such as positive airway pressure, supplemental oxygen and postnatal sepsis have also been found to be related to BPD,7 as occurred in our study.

Several studies have described patterns of disease that can usually be identified in the first weeks of life.29,30 First, there are infants with mild pulmonary disease at birth that recover gradually. Then, there are infants that experience early persistent pulmonary deterioration and usually require prolonged respiratory support starting from birth. Lastly, there are infants who recover from initial lung disease but later experience respiratory decompensation.7 This goes beyond what Northway described in 1967 in patients with BPD, then defined as the pulmonary disease developed by preterm infants subjected to prolonged mechanical ventilation.31 Today’s preterm infants do not have the same characteristics and are managed differently compared to the population in which BPD was first defined. Strategies used to improve lung development outcomes (used of synchronised ventilation modalities, standardization of lung volumes, increased use of NIV, early administration of surfactant and antenatal steroids) have resulted in the different population of patients we manage today.29 Every infant in our study was managed with these methods, as we obtained a consecutive sample of infants managed in the past decade. Thus, fewer than 4 in 10 patients received IMV during the hospital stay, and fewer than 2 in 10 required endotracheal intubation at birth. Notwithstanding, there is still a group of patients in whom the need of intubation and of IMV cannot be avoided.

When we analysed the relationship between IMV and BPD in the sample, we found that the need of IMV in the first days post birth was independently associated with the development of BPD, as previously described.32,33 Sixty percent of the patients that required intubation during stabilization in the delivery room developed BPD, and the percentage increased to 70% in those who required IMV in the first day of life. In contrast, only 14% of patients that were not intubated in the first day of life developed BPD. Similarly, nearly 9 out of 10 patients that needed IMV on day 3 post birth developed BPD and more than 50% moderate-severe BPD.

Our findings evinced another opportunity for improvement in the use of NIV for initial respiratory support. In our sample of very and extremely preterm infants, 8 in 10 received non-invasive respiratory support in the delivery room. In the group of patients that developed BPD, 60% had received CPAP/NIPPV at birth. However, this proportion decreased to 40% in the first 24 h post birth. Thus, there was a sizeable group of patients in who NIV failed in the first hours post birth and who eventually developed BPD. This was consistent with the findings in other case series in which CPAP failure was associated with an increase in morbidity, including BPD.34 Thorough investigation of these cases to establish NIV protocols and improve the respiratory management of these patients is an essential step that could help avoid IMV in a particularly important stage of disease development.33,35,36

Multiple previous studies have demonstrated the association between postnatal sepsis and the development of pulmonary inflammation.37–39 Some have found evidence suggestive of a causal relationship between postnatal infection, PDA and increased duration of IMV.37,39 All these factors are probably interrelated. However, in our study nosocomial sepsis remained an independent risk factor in the final logistic regression model. In fact, its presence was associated with a more than 9-fold increase in the risk of BPD as well as with the severity of BPD. Thus, in our population, nosocomial sepsis should be considered a risk factor in and of its own. Improving the rate of nosocomial infection through stricter protocols and limited use of invasive procedures is another of the preventive strategies that should be prioritised.

There are limitations to our study, including its retrospective design, performance in a single centre and small sample size, in addition to the inclusion of preterm infants born between 28 and 32 weeks of gestation, in who the baseline risk of BPD is low. The definition of BPD as oxygen dependence at 28 days post birth also carries intrinsic limitations, such as the discrepancy with the criterion established in the 2001 consensus (dependency for 28 days). This could hinder comparisons with the results of other studies. The values of the AUCs in the ROC curves of our models should be interpreted with caution, as we did not select variables for application at specific time points and we did not correct for the overestimation that may result from testing the model in the same patients from who it was derived. Our conclusions should be analysed with prudence and are only directly applicable to the population under study. On the other hand, the study was conducted in a homogeneous consecutive sample spanning 7 years during which the same protocols were applied and there were no significant changes in clinical management, which was akin to clinical practice in other facilities in our region.

In conclusion, the analysis of simple and objective data available in the first days post birth, such as GA at birth and the need of IMV in days 1 and 3 of life, can help identify a subset of patients at high risk of BPD in whom more aggressive management may be warranted. Our findings may provide the foundation to develop early-stage predictive models or to compare the predictive power of new tools using clinical data that can be obtained at the bedside.

Data will be made available on request.

This research did not receive any external funding.

The authors have no conflicts of interest to declare.