Evaluation of trends in Bronchopulmonary Dysplasia and Respiratory Support Practice for Very Low Birth Weight Infants: A Population-Based Cohort Study

Abstract

Objectives:

To characterize the incidence of bronchopulmonary dysplasia (BPD) over time and test the association of multi-level factors, including respiratory support, with the diagnosis of BPD.

Study design:

This population-based cohort study included 40,268 infants born between 22 and 32 weeks of gestation at hospitals in California between 2008 and 2017. BPD diagnosis was based on respiratory support at 36 weeks post-menstrual age. Tests for linear trend and multivariable logistic regression analyses were performed.

Results:

The rate of BPD was consistent year over year, and the mortality rate declined. The incidence of BPD was 23.5% for the overall cohort, and was 44.9% for less than 28 weeks gestational age and 45.2% for extremely low birth weight infants. For infants >26 weeks gestational age, the incidence of BPD for the most recent three years significantly decreased compared with vthe earlier three years (odds ratio (OR) 0.91). Invasive ventilation during delivery room resuscitation (OR 2.64) and after leaving the delivery room (OR 10.02) conferred the highest risk of BPD compared with oxygen or no respiratory support. Noninvasive ventilation as maximum respiratory support at 36 weeks increased by 20% over time.

Conclusions:

Marked changes to non-invasive support care have occurred without an overall decline in BPD rates. Further research, quality improvement, and strategies in addition to noninvasive respiratory support are needed for reduction of BPD.

Although morbidity and mortality has improved for very low birth weight infants in the United States; rates of bronchopulmonary dysplasia (BPD) have not significantly improved over the last decade.^{1, 2} Advances in respiratory therapy and nutritional care in preterm infants have held promise to reduce the prevalence of BPD. Ongoing contemporary reports of population-level rates can help to assess progress and gaps for further work.

BPD is the major cause of a prolonged length of stay in the neonatal intensive care unit (NICU)³ and is associated with poor cognitive and motor outcomes for premature infants.^{4, 5} Strategies to prevent BPD include both maternal and early postnatal interventions, such as antenatal steroids, surfactant therapy, and protective noninvasive ventilation,^6–8 but these existing strategies have had limited impact in reducing the burden of BPD.⁹ The challenges posed by BPD-related morbidity are consequently rising, as are the public health costs associated with prematurity and its comorbidities.¹⁰ Hence, there is ongoing need to seek innovative approaches and more effective therapies for BPD.

Avery et al suggested that NICU respiratory practices could influence the rate of chronic lung disease among preterm infants.¹¹ Multiple studies have suggested the positive effects of noninvasive respiratory support with continuous positive airway pressure (CPAP), as opposed to intubation for the prevention of death and BPD.^{12, 13} In meta-analysis, CPAP resulted in a small but clinically important reduction in BPD and the combined outcome of BPD and mortality.¹⁴ Indeed, multiple noninvasive respiratory support approaches have been adopted by many NICUs. However, there is a gap in knowledge of recent trends in BPD rates and respiratory support practices using population based cohorts.

The practice of using prolonged, potentially harmful, noninvasive respiratory support has also been questioned. One report from a 2005 cohort found that increased use of noninvasive nasal continuous positive airway pressure (CPAP) and supplemental oxygen was associated with increased BPD incidence. Adopting newer, noninvasive methods of respiratory support warrants further investigation of possible downstream clinical effects.¹⁵ BPD is a complex and multifactorial disease; a recent BPD workshop suggested novel methods for delivery room and early life interventions to prevent BPD⁷

In this study, we investigated the trends in BPD and tested the association of multi-level factors, including respiratory support, with the diagnosis of BPD.

Methods

Data were extracted from the California Perinatal Quality Care Collaborative (CPQCC) database, which included data collected across 132 units that accounted for the care of more than 95% of very low birth weight (VLBW) infants born in California during the study period. Among 177,926 infants assessed for eligibility from the CPQCC database between 2008 and 2017, we screened 49,333 VLBW infants who were born with a birth weight of 400–1500 g or a gestational age (GA) of 22–31 weeks without any severe congenital anomalies. After excluding 3,771 infants who died before 36 weeks postmenstrual age (PMA) and 3,586 infants who were below 21 or above 33 weeks GA, 40,268 infants were included in the final analysis. The infants were stratified by 4 gestational age groups (<26, 26–27, 28–29, 30–31 weeks) and 5 birth weight groups (401–750, 751–1000, 1001–1250, 1251–1500, <1500 g). The incidence of BPD from the earliest three years of the study period (2008–2010) was compared with the incidence from the most recent three years (2015–2017). The institutional review board at Stanford University approved the primary collection and analysis of the CPQCC data.

Trained data abstractors prospectively gathered maternal, delivery, and neonatal data based on definitions from the CPQCC operation manual, which uses standard definitions developed by the Vermont Oxford Network. BPD was defined as the receipt of supplemental oxygen at 36 weeks, or if the infant was discharged and never re-admitted, at 34 or 35 weeks PMA and on oxygen at the time of discharge. Infants discharged prior to 34 weeks PMA without supplemental oxygen were included in the at-risk population and were considered to not have BPD; infants using oxygen at that timepoint were excluded from the study. Additionally, if the GA or oxygen status were not reported, the infant was excluded. Inhaled nitric oxide (iNO) was defined as treatment for >4 hours, including both contiguous or additive treatment. Data on respiratory support at 36 weeks PMA was only available after 2011. Nasal intermittent mandatory ventilation (NIMV) included nasal or synchronized IMV or any other form of nonintubated assisted ventilation for greater than 4 hours. Nasal CPAP was defined as applying continuous positive airway pressure through the nose. High flow nasal cannula was defined as ≥ 1 liter per minute or more and low flow was defined as < 1 liter per minute. Maximum respiratory support post-delivery room was defined as support occurring at any time after leaving the delivery room or the initial resuscitation area.

Statistical analyses

Linear trends in outcomes and maximum respiratory support were assessed with Cochran-Armitage trend tests. At the hospital level, any NICU with an absolute decrease in BPD incidence >0 over 2008–2017 was designated as having any decrease, and we examined distributions of the number of NICUs with any decrease according to various hospital-level factors using chi-squared or T-tests. We tested for decreasing incidence of BPD over the most recent three years (2015–2017) compared with the earliest three years of the study period (2008–2010) according to infant age and weight groups using both Cochran-Armitage trend tests and unadjusted odds ratios to convey magnitude. To examine the association between maximum levels of respiratory support and BPD, we then constructed multivariable logistic regression models to estimate adjusted odds ratios (OR) of BPD with 95% confidence intervals (CIs). The multivariable logistic models were adjusted according to established CPQCC models, which includes the following covariates: California Children’s Services (CCS) level of NICU care,¹⁶ gestational age, gestational age (squared), small for gestational age, presence of congenital abnormality, and multiple birth. We calculated predicted probabilities of BPD according to respiratory support in the delivery room or post-delivery room using the least-squares means procedure, which produces the mean probability of BPD according to levels of respiratory support, given that all other covariates are set to their means. All analyses were performed using SAS v 9.4 (SAS Institute, Cary, North Carolina). A p-value < 0.05 was considered statistically significant.

Results

The incidence of BPD was 23.5% for the overall cohort. The proportion of BPD decreased with each week of increasing GA, with 59% of infants at <26 weeks GA with BPD, compared with 7% of infants >30 weeks. BPD incidence also decreased with increasing birth weight (Table I). More than 61% of extremely low birth weight (ELBW) infants, and 44.8% of infants below 28 weeks of gestation developed BPD. Moreover, 26% of male infants developed BPD, and 35.6% of infants with a 5-min Apgar score ≤ 7 developed BPD (Table 1).

Table 1.

Demographic characteristics

	No BPD n=30772 n (%)	BPD n=9496 n (%)
Male	15388 (74)	5429 (26)
Gestational age (wks), mean (SD)	28.8 (2.1)	26.5 (2.2)
<26	2492 (41)	3592 (59)
26–27	5575(65)	2974 (35)
28–29	10745 (84)	1992 (16)
≥ 30	11960 (93)	938 (7)
Small for gestational age	7725 (25)	1817 (19)
Birth weight, grams, mean (SD)	1188(253)	891(263)
401–750	2157 (39)	3397 (61)
751–1000	5959 (64)	3298 (36)
1001–1250	9386 (84)	1759 (16)
1251–1500	12096 (93)	925 (7)
> 1500	1162 (93)	84 (7)
Outborn	7767 (72)	2979 (28)
Race/Ethnicity Black	3959 (76)	1217 (24)
Hispanic	13830 (76)	4378 (24)
White	8195 (77)	2442 (23)
Asian/Pacific Islander	3837 (77)	1159 (23)
Native American	98 (79)	26 (21)
Other	693 (75)	235 (25)
1-min Apgar Score, mean (SD)	5.9 (2.3)	4.6 (2.4)
< 4	5275 (62)	3277 (38)
4–7	15382 (76)	4863 (24)
8–10	9957 (89)	1287 (11)
5-min Apgar Score, mean (SD)	7.8 (1.5)	6.9 (1.9)
< 4	861 (54)	728 (46)
4–7	8267 (65)	4356 (35)
8–10	21493 (83)	4346 (17)
Inhaled Nitric Oxide >4 hours	391 (32)	846 (68)
Surfactant	15224 (65)	8124 (35)
Maternal Chorioamnionitis	2101 (69)	923 (31)
Maternal diabetes	3865 (80)	987 (20)
Maternal hypertension	9666 (80)	2382 (20)
Bleeding/Previa	5358 (72)	2036 (28)
Preterm premature ROM	5861 (76)	1833 (24)

Values are number (%)

SD, standard deviation; ROM, rupture of membrane

In total, 57 hospitals showed decreasing BPD incidence from 2008 to 2017, and 63 hospitals did not. There were no significant differences in level of neonatal care, teaching hospital status, neonatologist coverage, or average number of hospital beds between hospitals with and without decreasing BPD incidence (Table 2).

Table 2.

Hospital Characteristics and decreasing BPD

Characteristic	Decreasing incidence of BPD	Increasing incidence of BPD	P-value
AAP Level
2	0 (0)	5 (7.9)	0.33
3	7 (12.3)	45 (71.4)
4	36 (63.2)	11 (17.5)
Unknown	14 (24.6)	2 (3.4)
Teaching hospital status
No	45 (80.4)	50 (80.6)	0.97
Yes	11 (19.6)	12 (19.4)
Neonatologist coverage
On call at home	31 (54.4)	42 (66.7)	0.17
On call at hospital	26 (45.6)	21 (33.3)
NICU beds, Mean (SD)	27.11 (17.4)	22.6 (16.8)	0.15

Values are number (%)

From 2008 to 2017, 40,268 eligible infants were included in CPQCC data registry. Overall, mortality significantly decreased in VLBW infants between 2008 and 2017 (p<.001 for trend), especially in infants with birth weight < 750 g, whose mortality rates decreased from 31.6 to 25.5 per 100 infants (P < .001 for trend). However, the overall BPD trend did not change (p=0.10 for trend) during this time (Figure 1; available at www.jpeds.com). Among infants above 26 weeks GA however, we observed a 2.3% decrease in the incidence of BPD for the most recent three years (16.6%) compared with the first three years (18.1%) (p=.002 for trend, unadjusted OR: 0.91, 95%CI 0.85–0.98). We found a similarly small but significant decrease in BPD incidence among infants greater than 750 g for the most recent three years (16.6%) compared with the first three years (18.5%) (p<.001 for trend, unadjusted OR 0.89, 95%CI 0.82–0.95) (Figure 2; available at www.jpeds.com).

Figure 1.

Trends of BPD and Mortality in California

Figure 2.

BPD incidence by birth weight group (A) and gestational age (B)

Trends for the maximum respiratory support provided in the delivery room, after the delivery room, and at 36 weeks post-menstrual age are shown in Figure 3 (available at www.jpeds.com). We found a decrease (46%) in invasive ventilation in the delivery room (p<.001 for trend), and non-invasive ventilation increased by 75% (p<.001 for trend) (Figure 3, A). Multivariable analyses showed that invasive ventilation in the delivery room conferred the highest odds of BPD incidence (OR 2.64, predicted probability 13.9%; Table 3).

Figure 3.

Trends of maximum respiratory support at Delivery room (A), post-delivery room (B) and 36 weeks of gestation (C) over the years

Table 3.

Risk of BPD according to maximum level of respiratory support and respiratory treatment.

	No. of BPD cases	Crude Rate of BPD (%)	Adjusted OR (95% CI)	Predicted Probability of BPD (%)
DR
No oxygen	167	5.27	1.0	6.17 (4.40–8.59)
Oxygen only	256	10.61	1.84 (1.49–2.27)	10.47 (7.66–14.18)
Non-invasive	2762	15.43	1.73 (1.46–2.05)	9.53 (7.10–12.69)
Invasive	6206	38.03	2.64 (2.22–3.13)	13.96 (10.54–18.27)
Post DR
No oxygen or oxygen only	45	1.68	1.00	1.71 (1.09–2.67)
Non-invasive	830	6.40	2.66 (1.96–3.62)	4.42 (3.15–6.18)
Invasive	8529	35.24	10.02 (7.40–13.58)	14.83 (10.97–19.75)
iNO use	841	68.65	3.79 (3.30–4.35)	31.60 (24.23–40.01)
Surfactant treatment	8043	34.84	2.88 (2.68–3.09)	14.94 (11.0–19.91)

BPD, bronchopulmonary dysplasia; DR, delivery room; iNO, inhaled nitric oxide; OR, odds ratio; CI, confidential interval

Rates of high frequency and conventional ventilation after the delivery room significantly decreased over time (p<.001 for trend) (Figure 3, B), and rates of NIMV and CPAP after the delivery room significantly increased (p<.001 for trend). The rate of infants for whom no oxygen was provided after leaving the delivery room also significantly decreased (p<.001 for trend). Invasive ventilation after leaving the delivery room resulted in a higher adjusted odds of BPD (OR 10.02) than receiving oxygen / no oxygen, with a 14.8% predicted probability of BPD (Table 3). Stratified by birthweight groups, the use of invasive ventilation decreased and non-invasive ventilation increased in all birthweight groups. However, infants who received no oxygen after the delivery room decreased only among infants <1500 g.

In terms of maximum respiratory support at 36 weeks PMA, we found a significant increase in invasive ventilation between 2011 and 2017 (p<.001 for trend), with the most substantial absolute increase occurring in ELBW infants. We also found an increase in non-invasive ventilation (p<.001 for trend). The number of infants without oxygen at 36 weeks decreased (p<.001), and this decrease was concentrated among ELBW infants (Figure 3, C,).

Discussion

Despite substantial shifts in respiratory support strategies and practice patterns, BPD rates have remained stable over the past decade. Risk for development of BPD is high for infants that receive higher degrees of respiratory support in the delivery room and the use of pulmonary surfactants. Because improved survival has not resulted in decreased BPD incidence, understanding respiratory support patterns in clinical settings is important for reducing the overall burden of pulmonary disease for preterm infants.

Other studies have reported similar rates of BPD in recent years. A multicenter study reported an overall BPD incidence of 42% in extremely preterm infants, with a range of 20–89% among different centers.¹⁷ In 2014, the Canadian Neonatal Network reported a 16–33% overall incidence of BPD in infants born at less than 33 weeks GA and noted that the incidence increased with decreasing GA.¹⁸ In our current study, we found that the incidence of BPD was 25.3% and 45.2% in VLBW and ELBW infants, respectively.

Although mortality significantly decreased among infants below 750 g, BPD incidence was unchanged or even slightly increased. In particular, infants with a birth weight of <750 g or <26 weeks GA had no significant change in BPD incidence in the most recent three years compared with the previous three years. These groups of infants should be targeted for quality improvement initiatives for reducing BPD incidence.

Several antenatal, perinatal, and postnatal factors contribute to BPD development.^19–21 Preterm infants often require assisted ventilation and/or supplemental oxygen after birth to ensure optimal gas exchange; however, this may also induce lung inflammation due to barotrauma and/or volutrauma and oxygen-free radical generation.¹⁹ Therefore, interventions which limit lung injury during resuscitation in the delivery room may help to prevent BPD development or reduce its severity. In our study, 41% of infants received invasive ventilation, 45% noninvasive ventilation, 6% supplemental oxygen, and 8% without oxygen in delivery room. Invasive ventilation (2% per year) and oxygen support (1% per year) decreased, and noninvasive ventilation increased (3% per year) over the 10-year period. A combination of respiratory management strategies and the underlying conditions resulting in the need to ventilate presumably contributed to increased risk of BPD, especially non-invasive (OR 1.73) and invasive ventilation (OR 2.64). Three large randomized trials for respiratory management after birth did not demonstrate significant reductions in death and BPD incidence rates at 36 weeks PMA for infants treated with CPAP compared with empiric intubation and mechanical ventilation.²² However, our findings support the hypothesis that minimizing respiratory support in the delivery room reduces lung inflammation and lung injury, thereby reducing the occurrence of BPD. This study also showed temporal change in the type of non-invasive support used, with increased use of CPAP and NIMV and decreased use of HFNC, which may reflect changes in practice following the results of a series of CPAP vs HFNC trials.

In 2010, CPQCC formed a delivery room QI Collaborative for the purpose of reducing hypothermia, delivery room intubations, and delivery room surfactant use, while increasing CPAP use as the initial ventilation strategy.²³ The Collaborative QI group was the only group with a reduction in BPD rate, and NICUs that worked on their own QI project in this area and non-participant groups had slight increases.²³ Thus, improved respiratory practice in the delivery room for infants <750 g or <26 weeks GA holds promise for further reducing the incidence of BPD. A recent BPD workshop held by the Eunice Kennedy Shriver National Institute of Child Health and Human Development suggested novel methods for delivery room and early life interventions to prevent BPD, including gentle ventilator approaches and the use of respiratory function monitors.⁷

Important risk factors for BPD include invasive mechanical ventilation, and prolonged ventilation.⁹ Several studies comparing high frequency ventilation with conventional mechanical ventilation have revealed a significant reduction in the incidence of death and BPD.²¹ We found that the total number of days of ventilation (OR 1.07) and the PMA at last ventilation (OR 1.14) increased the risk of BPD. Rates of high frequency ventilation and conventional ventilation markedly decreased in the delivery room and after NICU admission, potentially contributing to the reduction in BPD. Primary CPAP therapy reduced the risk of BPD and death compared with mechanical ventilation.²⁴

Adverse antenatal factors, such as chorioamnionitis, preeclampsia, pre-existing hypertensive disorders, gestational diabetes, and maternal obesity have been strongly associated with an increased risk of BPD.^25–27 A meta-analysis found that chorioamnionitis exposure is significantly associated with BPD (OR, 1.29).²⁸ In this study, gestational diabetes, pregnancy induced hypertension, and maternal chorioamnionitis were associated with a higher incidence of BPD. Also, small for gestational age status was associated with decreased incidence of BPD, which is contrary to other studies.²⁹

Several studies have explored how center-specific practices impact neonatal clinical outcomes.^{30, 31} In a 2015 analysis of CPQCC data, adjusted median BPD rates stratified according to the American Academy of Pediatrics neonatal level of care were 50.3%, 46.1%, and 47.7% for levels II, III, and IV, respectively, indicating variability across hospitals, even after accounting for patient case mix.³¹ Another study unexpectedly found that centers which used more CPAP did not have lower death or BPD incidence rates.³⁰ In the present study, 57 hospital showed decreased incidence rates of BPD over a 10-year period, and 63 hospitals had unchanged rates; we found no differences in level of care or average number of hospital beds between the hospitals with decreasing incidence of BPD and without.

This study should be viewed in light of its design. CPQCC does not collect data for BPD based on a physiological definition; instead, without data on the fraction of inhaled oxygen and other measures, it uses the revised BPD definition as oxygen at 36 weeks. In addition, there is a lack of information on institutional discharge criteria specifically for non-respiratory conditions. Moreover, although the CPQCC database contains a broad range of maternal and neonatal clinical variables, other unobserved key factors that could contribute to BPD incidence may exist that are not included in the database. As an observational study, we are limited in identifying causal relationships. Finally, although conducted in a single state, this study’s generalizability is supported by the population-based nature of the CPQCC data, and California’s diversity with respect to population, geography, and perinatal health care delivery models. California is also the most populous state in the United States.

In conclusion, despite changes to less invasive respiratory support in the care of VLBW infants, BPD remains a pressing issue. Quality improvement initiatives are needed to target high-risk groups, with strategies that go beyond increased use of noninvasive ventilation in the delivery room when appropriate, but also during the hospitalization.

List of abbreviations

AUC

area under the curve

BPD

bronchopulmonary dysplasia

CI

confidence interval

CPAP

continuous positive airway pressure

CPQCC

California Perinatal Quality Care Collaborative

GA

gestational age

iNO

inhaled nitric oxide

NICU

neonatal intensive care unit

OR

odds ratio

QI

quality improvement

ROC

receiver operating characteristic

VLBW

very low birth weight