Article Text
PDF
Abstract
Objective To evaluate the relationships among cerebral palsy (CP) phenotypes and bronchopulmonary dysplasia (BPD) severity and, in the process, to generate hypotheses regarding causal pathways linking BPD to CP.
Study design We studied 1047 infants born before the 28th week of gestation. Receipt of supplemental oxygen at 36 weeks postmenstrual age (PMA), with or without the need for mechanical ventilation (MV) at 36 weeks PMA, defined two levels of BPD. At 24 months, the children underwent neurologic examinations and CP diagnoses were made using an algorithm based on topographic localisation.
Results The 536 infants with BPD were at increased risk of all three CP phenotypes. In time-oriented multivariable analyses that adjusted for potential confounders, receipt of supplemental oxygen without MV at 36 weeks PMA (BPD) was not associated with increased risk of any CP phenotype. In contrast, BPD accompanied by MV at 36 weeks PMA (BPD/MV) was associated with a nearly sixfold increased risk of quadriparesis and a fourfold increased risk of diparesis.
Conclusions Combined treatment with both MV and supplemental oxygen at 36 weeks PMA strongly predicts the more common bilateral CP phenotypes. BPD without MV at 36 weeks PMA was not significantly associated with any form of CP.
Statistics from Altmetric.com
Introduction
Preterm infants with bronchopulmonary dysplasia (BPD) appear to be at increased risk for subsequent cerebral palsy (CP) in some studies,1,–,9 but not in others.10,–,14 Analysis of a BPD-CP association is complex for several reasons. First, BPD often follows exposures thought to influence CP risk, including extremely preterm birth, cerebral white matter damage,6 11 14,–,17 necrotizing enterocolitis,11 pneumothorax13 and prolonged mechanical ventilation (MV).9 13 14 In addition, a number of important changes in neonatal intensive care practices have been introduced over the past 15 years that might modify both BPD and CP risk. These include more widespread use of surfactant,11 12 antenatal corticosteroids14 and indomethacin;14 and increased intrapartum antibiotic prophylaxis accompanied by reduction in rates of early neonatal sepsis.17 Finally, the use of high dose postnatal steroids, which increased in the 1990s and has decreased since, is associated both with reduced severity of BPD and increased risk of CP.14 15 18 19
What is already known on this topic
Among surviving preterm infants, a diagnosis of bronchopulmonary dysplasia is associated with increased risk of later cerebral palsy.
What this study adds
▶ The association between cerebral palsy (CP) and bronchopulmonary dysplasia (BPD) varies with CP subtype and BPD severity.
▶ More severe BPD, defined as the need for both supplemental oxygen and mechanical ventilation at 36 weeks PMA, is associated with increased risk of quadriparesis and diparesis, but not hemiparesis.
In this study, we explored the relationship between BPD and CP, including CP phenotypes, while considering both potential shared antecedents as well as possible intermediaries in the causal pathway to CP.
Methods
The ELGAN study
The ELGAN study was designed to identify characteristics and exposures that increase the risk of structural and functional neurological disorders in extremely low gestational age newborns (ELGANs). During the years 2002–2004, 1506 infants delivered before 28 weeks gestation at 1 of 14 participating institutions were enrolled. The enrolment and consent processes were approved by the individual institutional review boards. For this report, we limited the sample to the 1047 children whose respiratory status at 36 weeks postmenstrual age (PMA) was known and who survived to 24 months post-term equivalent and had a complete neurological exam (figure 1).
Description of the study sample.
Demographic and pregnancy variables
After delivery, a trained research nurse interviewed each mother in her native language using a structured data collection form and following procedures specified in a study manual. The research nurse also reviewed the maternal medical record using a second structured data collection form. The medical record provided source information for clinical details and events following admission. The mother's report of her own characteristics and exposures, as well as the sequence of events leading to preterm delivery, were accepted as accurate, even when her medical record provided discrepant information.
The following clinical circumstances leading to each maternal admission and ultimately to preterm delivery were operationally defined as ‘initiators of preterm delivery’20: preterm labour, preterm pre-labour rupture of fetal membranes, pre-eclampsia, cervical insufficiency, placental abruption and fetal indications, including non-reassuring fetal testing, oligohydramnios, Doppler abnormalities of umbilical cord blood flow or severe intrauterine growth restriction based on clinical ultrasound testing.
A course of antenatal glucocorticoid was considered complete if the mother received two doses of betamethasone 24 h apart or if she received four doses of dexamethasone at 12 h intervals and delivered at least 48 h after the first dose of either medication.
Newborn variables
The gestational age estimates were based on a hierarchy of the quality of available information.20 The birth weight Z-score was defined as the number of SDs the infant's birth weight is above or below the median weight of infants at the same gestational age in a standard data set.21
Details were collected about the newborns physiological and laboratory status during the first 12 h to calculate a Score for Neonatal Acute Physiology (SNAP-II).22 In the first postnatal week, details about the newborn were collected daily. Thereafter, data were collected at weekly intervals through the first postnatal month (days 7, 14, 21, 28).
The diagnosis of BPD was assigned to children who received supplemental oxygen at 36 weeks PMA. Lacking detailed FiO2 data at 36 weeks gestation among those not ventilated, we defined two levels of BPD: receiving supplemental oxygen without ventilation (BPD) and with MV (BPD/MV) at 36 weeks PMA. The infant was classified as receiving MV if he or she was managed on mechanical support that included endotracheal intubation. MV included conventional ventilation in a variety of modes, as well as high frequency ventilation. Infants in room air on CPAP (n=5) at 36 weeks PMA were classified as having no BPD. At 36 weeks PMA, no infants were receiving MV without supplemental oxygen.
Infants were classified by their respiratory characteristics during the first two postnatal weeks as ‘persistently low FiO2’ (fraction of inspired oxygen consistently below 0.23), ‘pulmonary deterioration’ (FiO2 <0.23 on any day between days 3 and 7 and FiO2 >0.25 on day 14) and ‘early and persistent pulmonary dysfunction’ (FiO2 >0.25 on any day between days 3 and 7 and on day 14).23
Pneumothorax, pulmonary interstitial emphysema, pulmonary haemorrhage and patent ductus arteriosus were recorded if these diagnoses were noted in the medical record. The diagnoses of patent ductus arteriosus were separated into those made only by clinical signs and those confirmed by echocardiography. Necrotising enterocolitis was classified according to the modified Bell staging system.24
Ultrasound protocol scans
Routine cranial ultrasound scans were performed by technicians at all of the hospitals using digitised high frequency transducers (7.5 and 10 MHz). Ultrasound studies always included the six standard quasi-coronal views and five sagittal views using the anterior fontanel as the sonographic window.25 The three sets of protocol scans were defined by the postnatal day on which they were obtained (days 1–4, 5–14, 15+, using the third protocol scan closest in time to 36 weeks PMA). Details about reading procedures are presented elsewhere.26
24-Month developmental assessment
At 24 months corrected age, study infants returned for a developmental assessment, which included a neurological examination27 and an assessment for the Gross Motor Function Classification System (GMFCS) to assess the severity of the motor disability related to CP.28 Eighty-eight per cent (1047/1190) of the children surviving to 24 months corrected age underwent complete neurological examination.
Those who performed the neurological examinations received formal instruction and studied a manual, a data collection form and an instructional CD designed to minimise examiner variability; subsequently, they demonstrated acceptably low variability.27 The topographic diagnosis of CP (quadriparesis, diparesis or hemiparesis) was based on an algorithm using these data.26
Data analysis
We evaluated the association between BPD and CP with the view that, aside confounders, there are at least three potential explanations for the observed association: BPD causes CP; common antecedents contribute both to BPD and CP; and interventions for BPD increase the risk of CP. We explored the hypothesis that one or more CP phenotypes might be linked to BPD severity (ie, BPD without MV vs BPD with MV).
We began data analysis seeking characteristics and exposures that might alter our estimates of the strength of BPD-CP associations. Variables associated with both BPD and a CP diagnosis at p ≤0.30 were considered potential confounders and candidates for inclusion in logistic regression (LR) analyses.29
Because postnatal phenomena, such as the need for respiratory support, can be influenced by antepartum factors, and these relationships can vary over time,30 we created LR models in which risk factors were ordered in a temporal pattern, so that the earliest occurring predictors/covariates of a CP diagnosis were entered first and those that were significantly associated with the outcome of interest were not displaced by later occurring covariates.13 31,–,36 For the time-oriented regression models, we categorise sets of antecedents/covariates by the time they occurred or were identified. Each variable set is called an epoch. We used a step-down LR procedure seeking a parsimonious solution without interaction terms. We also fit maximum-likelihood multinomial LR models (also known as polychotomous LR) which simultaneously compared the children with each of the CP diagnoses to children who were not given a CP diagnosis.37 38
Results
In all, 120 children developed CP. The 536 children who had BPD were at greater risk of a CP diagnosis than the 511 children who were not oxygen dependent at 36 weeks PMA (table 1). Approximately half (n=64) of the infants with CP had quadriparesis and about a third (n=37) had diparesis. Only 19 children had hemiparesis, which limited our ability to draw inferences about this group. The risk of quadriparesis or an inability to walk, even with assistance (ie, GMFCS ≥2), was greater among the 17% (93/536) of infants with BPD/MV than among infants with BPD who were not on a ventilator. Children with CP tended to be ventilated and to receive supplemental oxygen longer than children who did not develop CP (figure 2). The differences were even more dramatic when children who could not walk (GMFCS ≥2) were compared to all the other children.
The cumulative probability of being ventilated or on oxygen each day, by cerebral palsy (CP) diagnosis and Gross Motor Classification System (GMFCS).
Children who were oxygen dependent at 36 postmenstrual weeks were at greater risk than their peers of a cerebral palsy (CP) diagnosis and Gross Motor Classification System (GMFCS) ≥2
Univariable analyses of potential antecedents and potential confounders
We examined variables that were associated both with BPD and CP as potential confounders of the BPD-CP association. Antenatal receipt of glucocorticoid for fetal lung maturation, caesarean section and initiator of preterm delivery were not associated with BPD or CP diagnosis (table 2). Infants of lower gestational age and sicker infants, however, were at increased risk of both entities. For example, infants born at 23–24 weeks gestational age were at a higher risk of BPD and of each of the three CP diagnoses than infants born during weeks 25–27. High values of the SNAP-II (table 3) also were associated with BPD and both quadriparesis and diparesis. Further, having an arterial blood gas drawn on postnatal days 7 and 14 was associated with an increased risk of BPD and both quadriparesis and hemiparesis.
Antenatal and perinatal variables associated with bronchopulmonary dysplasia (BPD) and cerebral palsy (CP) diagnoses
Postnatal variables associated with bronchopulmonary dysplasia (BPD) and cerebral palsy (CP) diagnosis
The set of cardiorespiratory factors associated with CP differed across CP phenotypes (table 3). Patent ductus arteriosus diagnosed by echocardiography was associated with BPD and all three CP phenotypes. Infants who developed pulmonary interstitial emphysema were at increased risk of BPD and of quadriparesis. Children who had ‘early and persistent pulmonary dysfunction’ or who developed ‘pulmonary deterioration’23 were at greater risk of BPD and quadriparesis or hemiparesis than were children who received consistently low FiO2 during the first two postnatal weeks.
Receipt of conventional or high frequency MV on the day of birth and days 14, 21 and 28 was associated with BPD and at least one CP phenotype (figure 3). Prior to 28 days, treatment with MV was associated only with an increased risk of quadriparesis. However, by day 28, treatment with MV was associated with each of the three CP phenotypes. BPD, with or without MV, was associated with an increased risk of quadriparesis and diparesis (figure 3 and table 1).
The percent of infants on mechanical ventilation on day of life 0, 7, 14, 21 and 28 and at 36 weeks postmenstrual age (PMA), by cerebral palsy (CP) diagnosis and Gross Motor Classification System (GMFCS).
Analyses of conditions associated with regional or systemic infection or inflammation yielded mixed results. Stage III or higher necrotising enterocolitis was associated with mildly increased risk of BPD and increased risks of each of the three CP diagnoses (table 3), while documented neonatal bacteraemia and tracheal infection (data not shown), as well as lower birth weight Z-score (table 2), were associated with BPD but not CP.
We also considered the possibility that treatments or medications might be associated with BPD and/or CP. Receipt of a methylxanthine (most often caffeine and rarely aminophylline or theophylline) was associated with reduced risk of BPD and quadriparesis (table 4). Fewer than 3% of infants received postnatal dexamethasone. On the other hand, the 11% of study infants treated with hydrocortisone were at increased risks of BPD and CP. Treatments for a patent ductus arteriosus and receipt of packed red blood cells or whole blood were associated with increased risks of BPD and each of the CP phenotypes.
Medication-related correlates of cerebral palsy (CP) phenotypes
Time-oriented multivariable analyses
Individual variables from the four epochs (antenatal, the first postnatal week, postnatal weeks 2–4 and 36 weeks PMA) were entered sequentially into the multivariable model. Each variable was selected based on observed associations, in bivariate analyses, with at least one of the three CP phenotypes.
The final model, which simultaneously adjusted for variables in all epochs, revealed different risk profiles for the three CP phenotypes. The risk of quadriparesis was increased with low gestational age, white race and public insurance from the first epoch and with the requirement for both supplemental oxygen and MV (BPD/MV) from the final epoch (table 5).
Multinomial risk model of cerebral palsy (CP)
The risk of diparesis was reduced for infants delivered by caesarean section, and increased with low gestational age, public insurance, pH in the lowest quartile for gestational age in two of the first three days and with BPD/MV.
Overall, BPD/MV was associated with a nearly sixfold elevated risk for quadriparesis and a fourfold risk for diparesis. Elevated risk for hemiparesis was associated with male sex and non-white race and reduced risk of hemiparesis was associated with public insurance. Neither BPD nor BPD/MV was associated with an increased risk of hemiparesis.
We also created time-oriented risk models that included, in the later postnatal epoch, two lesions seen on cranial ultrasound scans (ie, ventriculomegaly or echolucency) which are associated with neurodevelopmental delays39 (data not shown). In these models, BPD/MV was associated with a nearly 10-fold elevated risk for quadriparesis and nearly a fivefold risk for diparesis.
Discussion
Our study confirms that markers of pulmonary illness severity are associated with CP and documents that the associations differ among CP phenotypes.40 Children who, at 36 weeks PMA, received MV in addition to supplemental oxygen (ie, BPD/MV)41 were at substantially increased risks of quadriparesis and diparesis, but not hemiparesis. Not surprisingly, greater CP severity (GMFCS ≥2 or inability to walk) most closely aligns with quadriparesis, and, consequently, also is associated with BPD/MV. In contrast, after adjusting for other significant antecedents, supplemental oxygen without MV at 36 weeks PMA (BPD) was not significantly associated with any form of CP.
The most common CP phenotypes associated with BPD, quadriparesis and diparesis, are also the two most common forms of CP that occur in ELGANs.40 Not surprisingly, these bilateral forms of CP reflect diffuse, bilateral cerebral hemispheric disease, often indicated by neonatal cranial ultrasound or later MRI white matter parenchymal lesions and/or ventricular enlargement, which occur in more than 12% of ELGANs.40 The lack of a comparable significant association of hemiparesis with BPD may relate to two factors. First, relatively few children (n=19) have this CP phenotype, compared to the bilateral forms, which may make risk calculations unstable. Second, the neuroimaging findings associated with hemiparesis often differ from the bilateral white matter damage that is commonly associated with quadriparesis and diparesis, suggesting a different disease process. A frequent ultrasound finding associated with hemiparetic CP in ELGANs, periventricular haemorrhagic ultrasonographic abnormalities thought to represent infarction,42 are largely unilateral lesions, and only occur in about 2% of infants with very low birth weight.42
In 1990, Cooke reported results of a case–control study of the relationship among BPD and CP phenotypes.2 Using discriminant analysis, these investigators found an association between BPD (defined as the need for supplemental oxygen at 28 days) and CP phenotypes. The different definition of BPD and restriction to an overall BPD definition, rather than two levels of BPD severity, provide one potential explanation for the variation in results between their study and ours. Our study also differs in defining the study cohort by gestational age, rather than birth weight, thus reducing over representation of growth restricted infants. The ELGAN Study also used standardised examination methods and an algorithm in order to determine CP phenotypes.
More contemporary reports suggest that infants who receive supplemental oxygen at 36 weeks PMA are at increased risk for global motor impairment43 44 and infants with severe BPD, based on the definition proposed by the National Institutes of Health (NIH) BPD Consensus Conference, are more likely to score low at 24 months corrected age on the Psychomotor Development Index component of the Bayley exam36 and to meet criteria for CP by neurological examination.44 Other investigators have reported increased risk of CP with failure of a mother to receive antenatal steroids,45 increasing duration of ventilation7 13 46 47 or the occurrence of pneumothorax.13 46
The associations among BPD, CP, antecedent risk factors and mediators in the causal pathways are complex, and might call for new conceptual and analytic approaches. For example, using an approach based on directed acyclic graphs, Gagliardi et al found that BPD was closely linked to intracranial white matter damage, an antecedent of CP, however the association was mainly explained by shared risk factors and causal pathways.48
We, too, represented the association between BPD and CP49 50 as causal directed acyclic graphs, envisioning four potential patterns of association among BPD, CP and other risk factors (figure 4).51 First, BPD might cause brain injury either through associated physiological effects or via treatment(s) for BPD (figure 4A). Second, risk factors for BPD might co-occur in temporal proximity to risk factors for CP, representing concomitant, but independent causal pathways (figure 4B). Finally, BPD and CP might share a common antecedent. In one example of this theme, the antecedent is identified (figure 4C). In another, an unknown or unidentified risk factor occurs concomitantly with a factor that is erroneously thought to be in the causal path (figure 4D). Because our data do not permit us to distinguish among these possibilities, the observed risks of quadriparesis and diparesis associated with more severe BPD could result from a causal relationship, shared antecedents, temporal coincidence or all three potential pathways.
Hypothetical models that might explain the associations among risk factors, bronchopulmonary dysplasia (BPD), and cerebral palsy (CP). CP might represent a ‘downstream’ consequence of BPD in which the risk factor is an intermediary (4A). Independent risk factors leading to BPD and CP might occur in synchrony (4B). CP and BPD might share one or more risk factors (4C). Finally, a ‘risk factor’ might act as a surrogate for causal factors, common vulnerabilities, or more critical predictors of BPD and CP (4D).
We offer an example of each of these hypothetical models to demonstrate their plausibility.
The first is for the model characterised by ‘processes leading to BPD contribute to CP’ (figure 4A). Pulmonary volutrauma, barotrauma, oxygen toxicity and local pulmonary infection induce inflammation that predisposes to lung injury.52,–,55 In this model, transition from local inflammation to a BPD-induced systemic inflammatory response56 increases the likelihood of associated brain injury and resultant CP.2 55 57,–,62 Alternatively, systemic inflammation might damage both lung and brain.
A variant of the ‘processes leading to BPD contribute to CP’ model (figure 4A) invokes the role of therapy. For example, at high doses, postnatal corticosteroids given to reduce the risk of BPD are associated with subsequent brain damage.19 63,–,65 In our cohort, however, the rate and doses of postnatal dexamethasone treatment were low.
The ‘temporal proximity’ model postulates independent causal pathways for BPD and CP (figure 4B). Underlying this model is the assumption that the lung and brain are especially vulnerable to perturbation at extremely preterm gestational ages. For example, in the critically ill newborn, lung structural immaturity and deficiencies of surfactant and antioxidants predispose to local pulmonary injury via barotrauma and oxygen toxicity leading to subsequent BPD, yet these factors might play no role in CP pathogenesis. Rather, simultaneously occurring factors, such as infection, might promote brain injury leading to CP.
Infection/inflammation can both exacerbate lung disease66,–,69 and predispose to cerebral white matter damage (figure 4C).70 71 Although there have been few investigations of the links between infection/inflammation and CP phenotype, one study found an especially strong association between clinical chorioamnionitis and diplegic CP.59
Developmental regulation of factors such as antioxidants or thyroid hormone7 34 72,–,74 also can serve as an example of the ‘common antecedents’ model of BPD and CP causation (figure 4C). In part because of their limited antioxidant capabilities, preterm infants appear to be more prone to oxidative stress than infants born at term.75,–,77 This has prompted some to suggest that several of the disorders that occur preferentially in preterm newborns, including BPD,78 cerebral white matter damage79 80 and CP,81 might be explained on the basis of oxidative stress and related inflammation.82 An alternative explanation is that biologically regulated factors serve simply as surrogates for other factors that modify the risk of BPD and CP (figure 4D).
BPD is diagnosed at 36 weeks PMA and CP is most often detected years later. Pathogenic mechanisms for the two disorders, however, substantially precede their manifestation. The temporal relationship of the diagnoses leads to the assumption that, if the disorders are causally related, it is BPD that predisposes to CP. Detection of brainstem lesions among infants with neonatal white matter disease,83 84 already at increased risk of CP, raises the possibility that early neonatal brain injury might also contribute to BPD through apnoea that prolongs the need for MV. There is, however, no evidence that apnoea predicts increased risk of CP85 and, in most cases, apnoea of prematurity is effectively treated with caffeine, a treatment that appears to ameliorate CP risk.86 Furthermore, brain damage severe enough to lead to prolonged respiratory support is at least as likely to be associated with subsequent CP as is consequently occurring BPD.
A major strength of our study is the large cohort that enabled us to evaluate a substantial number of cases of BPD and CP and permitted investigation of the relationship among categories of BPD severity and CP phenotypes. Although each of the CP diagnoses occurred among fewer than 8% of children, the magnitude of the increased risk of a CP phenotype with a BPD entity (ORs between 4.2 and 5.7) was sufficient to achieve statistical significance.
Information about potential antecedents of BPD and CP diagnoses was collected prospectively, as were data about characteristics and exposures that might confound the association between BPD and CP. To maximise data quality, ultrasound scans were read and reported in a specified format designed for this study and two readers had to agree that an abnormality was present.26 Further, the neurological assessments followed formal training of all examiners and were highly structured, completed by examiners who were unaware of the infants' medical histories, and monitored to ensure consistent standards and high inter-rater reliability.27 A standard algorithm was used to classify each affected infant's CP phenotype.26
This study has the potential limitations of all observational studies. Specifically, our study results might be explained by failing to account for one or more unrecognised confounding factors.87 Further, although based on high quality prospectively collected data, in applying a two-level BPD definition (supplemental oxygen with vs without mechanical ventilation) we chose a definition that was neither as nuanced as the categories proposed by the NIH Workshop on BPD41 44 nor as rigorous as one defined by a physiological test for BPD.88 Finally, analyses of the hemiparesis subgroup were limited somewhat by the small size of this group. In the multivariable model of hemiparesis, the diminished effect of lower gestational age, white race, lower socioeconomic status (as indicated by public insurance) and the lack of an association with BPD or BPD/MV suggests the relationship with BPD differs from those observed for quadriparesis and diparesis. Accounting for only 16% of all CP (19 of 120), however, hemiparesis was the least common CP phenotype among our study population. Although it appeared to be aetiologically distinct from quadriparesis and diparesis, we were unable to conduct a detailed exploration of all potential aspects.
Conclusions
We conclude that the associations between BPD and CP are attenuated when taking into account variables associated with prematurity and correlates of illness severity. The residual association between BPD and CP varies by BPD severity and CP phenotype. BPD/MV was strongly associated with an increased risk of quadriparesis and diparesis. Although causal inferences cannot be made on the basis of this observational study, future investigations of preventive therapies or pathophysiological mechanisms underlying the BPD-CP relationship should take into account the likelihood of variability in antecedent relationships among CP phenotypes.
Acknowledgments
We gratefully acknowledge the participation and contributions of the ELGAN study subjects and their families, as well as those of our colleagues. We thank Elizabeth Horner and Alison Clapp for assistance with clerical and literature elements of the manuscript. Finally, we would like to thank the additional ELGAN Study collaborators who made this report possible:
Children's Hospital Boston, Boston, MA: Haim Bassan, Samantha Butler, Adré Duplessis, Cecil Hahn, Catherine Limperopoulos, Omar Khwaja, Janet S Soul; Baystate Medical Center, Springfield, MA: Bhavesh Shah, Herbert Gilmore, Susan McQuiston; Beth Israel Deaconess Medical Center, Boston, MA: Camilia R Martin; Massachusetts General Hospital, Boston, MA: Robert M Insoft, Kalpathy Krishnamoorthy; Floating Hospital for Children at Tufts Medical Center, Boston, MA: Cynthia Cole, John M Fiascone, Paige T Church, Cecelia Keller, Karen J Miller; U Mass Memorial Health Care, Worcester, MA: Francis Bednarek, Robin Adair, Richard Bream, Alice Miller, Albert Scheiner, Christy Stine; Yale University School of Medicine, New Haven, CT: Nancy Close, Elaine Romano, Joanne Williams; Wake Forest University Baptist Medical Center and Forsyth Medical Center, Winston-Salem, NC: Deborah Allred, Robert Dillard, Don Goldstein, Deborah Hiatt, Gail Hounshell, Ellen Waldrep, Lisa Washburn, Cherrie D Welch; University Health Systems of Eastern Carolina, Greenville, NC: Stephen C Engelke, Sharon Buckwald, Rebecca Helms, Kathyrn Kerkering, Scott S MacGilvray, Peter Resnik; North Carolina Children's Hospital, Chapel Hill, NC: Lisa Bostic, Diane Marshall, Kristi Milowic, Janice Wereszczak; Helen DeVos Children's Hospital, Grand Rapids, MI: Mariel Poortenga, Wendy Burdo-Hartman, Lynn Fagerman, Kim Lohr, Steve Pastyrnak, Dinah Sutton; Sparrow Hospital, Lansing, MI: Nicholas Olomu, Victoria J Caine, Joan Price; Michigan State University, East Lansing, MI: Nigel Paneth, Padmani Karna; University of Chicago Medical Center, Chicago, IL: Leslie Caldarelli, Sunila E O'Connor, Michael Msall, Susan Plesha-Troyke; William Beaumont Hospital, Royal Oak, MI: Daniel Batton, Karen Brooklier, Beth Kring, Melisa J Oca, Katherine M Solomon.
References
Footnotes
-
Funding This study was supported by a cooperative agreement with the National Institute of Neurological Diseases and Stroke (5U01NS040069-05) and a Mental Retardation and Developmental Disabilities Research Center grant from the National Institute of Child Health and Human Development (NIH-P30-HD-18655). CB was partially supported by the Thrasher Research Fund. LJVM was partially supported by the National Institutes of Heart, Lung, and Blood (1 P01 HL 67669-01).
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval This study was conducted with the approval of all 14 sites.
-
Provenance and peer review Not commissioned; externally peer reviewed.