Abstract

Introduction

Bronchopulmonary dysplasia (BPD) is a serious morbidity among preterm infants, with a reported incidence of 42% among infants 501-750 grams birth weight and 25% among those with 751-1000 grams birth weight (1). Traditional BPD, defined as supplemental oxygen (O₂) administration at 36 weeks postmenstrual age was diagnosed in 42% of a recent cohort of preterm infants 22 to 28 weeks gestation from the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network (NICHD NRN) who survived to 36 weeks postmenstrual age (2). BPD is associated with an increased risk of neurodevelopmental impairment with mental, psychomotor and language delays in early childhood (3, 4). At school age, children with BPD have greater risk of growth retardation and academic difficulties (5, 6).

The diagnosis of BPD has evolved in recent years from that based solely on duration of O₂ administration to one incorporating a standardized assessment of O₂ need. The traditional definition of BPD is based on O₂ administration at a postnatal age of 28 days or postmenstrual age of 36 weeks. The National Institutes of Health (NIH) consensus definition of BPD includes a classification of severity (mild, moderate and severe) based on O₂ use at 28 days and at 36 weeks postmenstrual age or discharge home for preterm infants < 32 weeks gestation (7). The predictive validity of the NIH consensus definition for 18 month outcomes has been demonstrated by Ehrenkranz and colleagues in the NICHD NRN in a large cohort of extremely low birth weight (ELBW) infants born between 1995 and 1999 (8). The incidence of any neurodevelopmental impairment, cerebral palsy, mental and psychomotor developmental indices < 70 on the Bayley Scales of Infant Development II and blindness and hearing impairment significantly increased as the severity of BPD increased. In another study, compared with children with mild or moderate BPD, those with severe BPD perform poorly on IQ tests at 3 years and had lower performance IQs and poorer perceptual organization at 8 years of age (9). In both these studies, moderate BPD was based on FiO₂ < 30% at 36 weeks postmenstrual age, an imperfect definition due to disparate individual and center practices related to target O₂ saturations and thresholds for O₂ therapy.

The physiologic definition of BPD, developed by Walsh et al, is based on a standardized assessment of O₂ saturation during a timed stepwise reduction of administered O₂ to room air, with clearly defined criteria for ‘passing’ or ‘failing’ the challenge (10). It has been demonstrated to be feasible, safe and reliable and to reduce the reported rate of BPD in individual centers of the NICHD NRN by a mean of 10% (11). The relationship between BPD by the physiologic definition and early childhood growth and neurodevelopmental outcomes has not been previously investigated. In this prospective observational cohort study from the NICHD NRN, we compared the growth and neurodevelopmental outcomes at 18-22 months corrected age of a recent cohort of ELBW (birth weights 401-1000 grams) infants with and without physiologic BPD. We further examined the association between BPD by the physiologic definition and cognitive impairment on the Bayley Scales III. Our hypothesis was that preterm ELBW infants with BPD by the physiologic definition would have a higher rate of adverse growth and neurodevelopmental outcomes at 18-22 months corrected age, compared to those without BPD by the physiologic definition and that BPD would be an independent predictor of cognitive impairment.

Material and Methods

This was a secondary analysis of data collected prospectively as part of the NICHD NRN very low birth weight registry (Generic Database) and Follow-up Study. Institutional Review Boards at each participating center approved both the database and follow-up studies.

Results

A total of 2,716 extremely low birth weight (ELBW) infants were admitted to the centers of the NICHD NRN between January 1, 2006 and June 30, 2007, 2065 of whom were eligible for the follow-up study. Of these, 1,189 infants were in the hospital at 36 weeks (725 had died by 36 weeks, 89 had transferred, and 62 had been discharged). Figure 1 is a flowchart describing the study cohort.

Open in a separate window

Figure 1

Flowchart for cohort and outcomes

Follow up outcomes

Seventy four (6.4%) infants died between 36 weeks postmenstrual age and 18-22 months corrected age. The vast majority of deaths (86.5%) occurred in those with BPD. Data at 18-22 months corrected age were available for 963 (88.7%) survivors, 92% and 88% in the groups with and without BPD. Growth and neurodevelopmental outcomes were compared between groups of surviving infants with and without physiologic BPD (Table 3). The mean weight and head circumference at follow up was significantly lower in those with BPD. A significantly higher proportion of infants with BPD had weight, length and head circumference < 10^th centiles at the follow-up visit. Moderate to severe cerebral palsy (7.0 vs. 2.1%) and spastic diplegia (7.8 vs.4.1%) and spastic quadriplegia (3.9 vs. 0.9%) phenotypes were significantly more frequent in those with BPD as were cognitive score< 70 (12.8 vs. 4.6%) and language score < 70 (24.2 vs. 12.3%), compared to those without BPD. A comparison of neurodevelopmental outcomes between the small numbers of infants who passed (n=49) and failed (n=60) the room air challenge revealed no significant differences (table 4).

Table 3

Follow-up outcomes of infants at 18-22 months corrected age.

Follow-up Outcome	BPD, N=484	No BPD, N=479	p-value
Maternal education,			0.14
highest level <12 y, %	23.1	23.8
HS diploma or greater, %	72.5	74.1
Medicaid or uninsured, %	62.8	67.2	0.17
Weight, kg, mean (SD)	10.5 (1.5)	10.7 (1.5)	0.01
Length, cm, mean (SD)	80.9 (4.2)	81.3 (5.4)	0.15
Head CF, cm, mean (SD)	46.7 (2.0)	47.0 (2.0)	0.02
Weight < 10th percentile, %	49.8	39.4	0.002
Length < 10th percentile, %	37.7	25.4	<.0001
Head CF < 10th centile., %	30.4	21.3	0.002
Blind1, %	1.9	0.6	0.15
Deaf2, %	2.3	0.6	0.06
Cerebral palsy, moderate.-severe., %	7.0	2.1	0.0004
Spastic diplegia, %	7.8	4.1	0.03
Spastic hemiplegia, %	2.2	1.3	0.45
Spastic quadriplegia, %	3.9	0.9	0.005
Cognitive score3
<70, %	12.8	4.6	<.0001
70-85, %	36.9	29.4
>85, %	50.3	66.0
Language score 4
<70, %	24.2	12.3	<.0001
70-85, %	33.8	27.8
>85, %	42.1	59.9

Open in a separate window

¹Blindness is defined as no useful function in either eye, or, possibly, blind with some functional vision (legally blind).

²Deafness is defined as requiring hearing aids in both ears.

³The cognitive composite scores from the Bayleys III exam administered at follow up.

⁴The language composite scores from the Bayleys III exam administered at follow up.

Table 4

Follow-up outcomes of infants who underwent the room air challenge at 18-22 months corrected age.

Outcome	Room air challenge, N=109
	Failed N=60	Passed N=49	p-value
Maternal education, highest level
<12 y, %	16.7	20.4	0.11
HS diploma or greater, %	75.0	79.6
Medicaid or uninsured, %	63.3	52.1	0.33
Weight, kg, mean (SD)	10.6 (1.26)	10.8 (1.64)	0.81
Length, cm, mean (SD)*	80.6 (4.17)	81.1 (4.28)	0.60
Head CF, cm, mean (SD)	47.0 (1.53)	47.2 (1.85)	0.88
Weight < 10th percentile, %	41.7	40.8	1.00
Length < 10th percentile, %	38.3	22.5	0.12
Head CF < 10th centile., %	26.7	14.3	0.18
Blind1, %	0.0	0.0	NA
Deaf2, %	0.0	0.0	NA
Cerebral palsy, mod.-sev., %	3.3	0.0	0.50
Spastic diplegia, %	6.9	2.2	0.38
Spastic hemiplegia, %	1.7	0.0	1.00
Spastic quadriplegia, %	1.7	0.0	1.00
Cognitive score3
<70, %	1.7	4.1
70-85, %	43.3	26.5	0.14
>85, %	55.0	69.4
Language score4
<70, %	16.7	10.4
70-85, %	33.3	29.2	0.49
>85, %	50.0	60.4

Open in a separate window

^*Does not include outlier data from one Passed infant.

¹Blindness is defined as no useful function in either eye, or, possibly, blind with some functional vision (legally blind).

²Deafness is defined as requiring hearing aids in both ears.

³The cognitive composite scores from the Bayleys III exam administered at follow up.

⁴The language composite scores from the Bayleys III exam administered at follow up. CF circumference

We further explored the association of BPD with cognitive impairment in a multivariate logistic regression model. Table 5 presents the adjusted odds ratios (with 95% CI) of the risk of cognitive impairment from logistic regression models. Physiologic BPD (adjusted OR 2.41; 95% CI 1.40-4.13) was independently associated with cognitive impairment. Other factors that were independently associated with adverse outcomes at 18 -22 months age included severe ICH or PVL, surgical NEC and small for gestation status.

Table 5

Adjusted odds ratios of the risk of Cognitive Score<70¹ from multivariate logistic regression, modeled with predictors and physiologic BPD.

Predictor variable	Odds Ratio	95% CI
Gestational age2	0.91	(0.76, 1.08)
Male gender	1.39	(0.86, 2.24)
Small for gestational age	2.60	(1.23, 5.50)
Maternal education
Unknown vs HS	1.79	(0.62, 5.15)
diploma+	1.41	(0.81, 2.43)
< HS vs HS diploma+
Surgical NEC	3.35	(1.42, 7.91)
IVH or PVL	3.97	(2.40, 6.55)
Physiologic BPD	2.41	(1.40, 4.13)

Open in a separate window

¹Cognitive composite score from the Bayley III exam.

²Change in odds of the outcome with each additional week of gestational age.

Discussion

This is the first investigation of the relationship between BPD determined by the physiologic definition and growth and neurodevelopmental outcomes using Bayley Scales III. Our cohort comprised a select subset of ELBW (< 1000 grams) infants, the majority of whom were < 27 weeks gestation and still hospitalized at 36 weeks postmenstrual age. We chose these restrictive inclusion criteria in order to obtain physiologic BPD data, using the standardized O₂ reduction challenge for infants with effective FiO₂ < 0.3. The relatively high rates of BPD in our cohort of extremely low gestation infants reflect our inclusion criteria. About 10% of our cohort underwent the room air challenge, similar to 14% in the original study on the impact of the physiologic definition of BPD; an identical 44% passed the challenge (11). The proportion of infants who died (6.4%) between 36 weeks and 18 months is consistent with previous reported rates between 2.2 and 11.2% (12, 13). In our data, death beyond 36 weeks postmenstrual age was overwhelmingly more frequent in those with BPD.

BPD is a recognized antecedent of cerebral palsy and adverse motor outcomes. Skidmore et al showed that cerebral palsy occurs more frequently (15% vs. 3-4%) in very low birth weight infants with BPD (defined as oxygen dependence at 28 days or beyond) (14). BPD by the traditional definition (supplemental oxygen at 36 weeks or discharge home) was one of the identified significant risk factors for cerebral palsy, neurodevelopmental impairment and psychomotor developmental index (PDI) < 70 at 18-22 months corrected age among 839 infants < 25 weeks gestation in the NRN (15). Singer and colleagues demonstrated that BPD was a significant, independent predictor of poorer motor outcome at 3 years of age after controlling for confounding variables and was associated with a 10-12 point decrement in PDI scores (16). In the more recent Extremely Low Gestational Age Newborn (ELGAN) study cohort, BPD accompanied by mechanical ventilation at 36 weeks postmenstrual age was associated with 6-fold increased risk of quadriparesis and 4-fold increased risk of diparesis (21). Our findings of a significantly higher rate of moderate or severe cerebral palsy, spastic diplegia and quadriplegia in those with BPD by the physiologic definition are consistent with these data.

Data on the association between BPD and cognitive impairment are conflicting. In one previous study, significantly higher rates of specific receptive language impairment were identified in VLBW infants with BPD compared to those without BPD (17). In the EPICure study, preterm infants ≤ 25 weeks gestation who were receiving supplemental oxygen therapy at 36 weeks postmenstrual age, when compared to those breathing room air had lower mean MDI score at 30 months corrected age after adjustment for the main predictor variables (18). In a secondary analysis of the trial of indomethacin prophylaxis in preterm infants, Schmidt and colleagues found that, among ELBW infants who survived to a postmenstrual age of 36 weeks, traditional BPD was independently associated (OR 2.4; 95% CI 1.8-3.2) with a poor 18-month outcome, defined as death or any of: cerebral palsy, cognitive delay, hearing loss requiring amplification, or bilateral blindness (13). A significant negative linear relationship between the “severity of BPD” and cognitive outcomes in infancy has been previously demonstrated (8, 19). In a recent analysis from the ELGAN study, Laughon and colleagues examined the association between chronic lung disease (defined as O₂ administration at 36 weeks postmenstrual age) and developmental delay at 2 years of age (20). The lowest risk of MDI or PDI < 55 was in those without chronic lung disease but chronic lung disease, without or with mechanical ventilation, was not associated with the risk of a low MDI. There are several possible reasons for this discordance between our results and the ELGAN study. In the ELGAN analysis, infants who were not able to walk independently and had moderate or severe gross motor impairments were excluded. Since cerebral palsy may be associated with cognitive impairment, it is possible that the association between cognitive impairment and BPD was underestimated. In addition, cognitive impairment was defined as MDI < 55 using Bayley II scales. The physiologic definition was used only in our cohort to define BPD. The mechanisms proposed for adverse neurodevelopment of infants with BPD include postnatal corticosteroid use impacting brain growth and recurrent episodes of hypoxia which may affect neuronal organization, myelination and cellular apoptosis (18). In addition, it is possible that antecedents of BPD are risk factors for delayed development (20).

Although a substantial body of work has examined the association between BPD and later outcomes, this is the first investigation in which BPD was rigorously defined based on performance of the room air challenge. The physiologic definition has been previously shown to obviate variability in oxygen use and provide a more uniform outcome for clinical comparisons between centers (22). BPD by the physiological definition is now being used in preference to the traditional definition in major randomized controlled trials (23). The present analysis supports the predictive validity of the physiologic definition of BPD for adverse neurodevelopmental outcomes at 18-22 months corrected age. Our data were prospectively collected from a large recent cohort when non-invasive modes of ventilation and surfactant were common and postnatal steroids were used selectively. The other major strength of the data is that neurodevelopmental assessment included structured neurologic examinations by certified examiners and the Bayley Scales of Infant Development III for the first time. The Bayley Scales III separately measure cognitive and language scores and have been recently shown to have significantly improved positive predictive value for low IQ at 4 years, compared to Bayley Scales II (0.87 vs. 0.29), with comparable predictive value for normal IQ (0.93 vs. 0.95) among extremely preterm infants (Bode MM, Mettelman BB, Maslak SE, Gross SJ. Predictive validity of the Bayley Scales of Infant Development 3^rd edition at age 24 months to predict preschool cognitive function in children born at ≤ 30 weeks gestation. EPAS 20112300.2).

The rates of cerebral palsy in the current data set were strikingly low (7% in those with BPD and 2.1% in those without BPD). Among infants with moderate and severe BPD born between 1995 and 1999 in any of the NRN sites, the reported rates of cerebral palsy were 17 and 26.8% respectively and in another large multicenter trial in the late 90s, rates in the groups with and without BPD were 17 and 9% (8, 13). The ELGAN study from a 2002-2004 cohort reported a cerebral palsy rate of 11.5% overall (21). Whether our data reflect improved outcomes over the past decade remains unclear. The proportion of infants with Bayley Scales II MDI < 70 in previous studies have ranged between 35.1 and 49.8% in infants with moderate and severe BPD and 19% in those without BPD (8, 13). The inherent differences between the Bayley Scales II and III, however, preclude direct comparisons with previous data from similar populations. The technical report of the Bayley-III states that mean scores on the Bayley–III are approximately 7 points higher than the Bayley–II MDI and PDI scores (24).

Our study has certain limitations. Our cohort comprised a select group of infants in hospital at 36 weeks postmenstrual age. Despite use of the room air challenge and the ‘physiologic’ diagnosis of BPD, practice variations in home oxygen use may have confounded subsequent outcomes. We did not collect data on the precise indication for supplemental oxygen use at 36 weeks or the timing of discontinuation of O₂. The numbers of infants who underwent the room air challenge were relatively small, and therefore, statistical differences in 18 month outcomes between those who passed and failed were not detected. Our focus was on neurodevelopmental outcomes; post-discharge pulmonary morbidities, potential modulators of developmental outcomes, were not examined. Finally, our results are confined to 18-22 months corrected age, a relatively early developmental stage.

We conclude that rates of adverse neurodevelopmental outcomes in early childhood were significantly higher in those with BPD. BPD by the physiologic definition was independently associated with cognitive impairment using the Bayley Scales III. These data on the prognostic importance of the physiologic BPD suggest that performing the room air challenge in infants on low supplemental oxygen at 36 weeks postmenstrual age may have implications for parental counseling and may allow for targeted post-discharge surveillance and early intervention.

Abbreviations

Footnotes

References