Introduction

Neurodevelopmental impairment (NDI) is among the most common outcomes reported in studies of neonatal prognosis and therapy. Guidelines recommend discussing NDI in antenatal discussions about what to expect if an infant survives [1]. The outcome of NDI has also become a standard primary endpoint of neonatal clinical trials [2]. Although definitions vary among studies, NDI is commonly defined as having one or more of the following: cerebral palsy, blindness, deafness, or a low score on an assessment of cognitive, language, or motor development (e.g., the Bayley Scales of Infant and Toddler Development) [3].

Some parents, researchers, and clinicians have questioned whether NDI should be extensively used as the primary outcome of prognostic and therapeutic research and whether it adequately represents the concerns of patients, families, and society [4,5,6,7]. Diagnoses such as cerebral palsy and developmental scores may not reflect the day-to-day impact of premature birth in areas such as the need for medical care (e.g., appointments, surgeries, rehospitalizations, and medicines) and daily functioning.

The Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Neonatal Research Network (NRN) collects detailed follow-up outcome data for children born extremely preterm (22–26 weeks’ gestation) at participating NRN centers and so provides a resource for describing the early lives of families of children with and without NDI. The purpose of this study was to describe concurrent outcomes of children born extremely preterm with and without diagnoses of NDI in order to guide clinical care and future research.

Methods

Population

This study included children born at 22 0/7 to 26 6/7 weeks’ gestation between 5/1/2006 and 6/30/2012 in 21 US academic centers participating in the NRN. The end date for the study was defined by the last births with follow-up examinations taking place at 18–22 months’ corrected age (for infants born after 7/1/2012, the timing of NRN follow-up assessments was changed from 18–22 months to 22–26 months’ corrected age). Infants with major congenital anomalies were excluded from the analysis, as factors besides prematurity may impact their outcomes. The institutional review board at each participating site approved NRN in-hospital and follow-up protocols.

Data collection

Trained research personnel at each center obtained data for all liveborn extremely preterm infants using previously described protocols [8, 9]. Gestational age at birth was defined by the best obstetric estimate. Small for gestational age was defined as <10th percentile by gestational age at birth using sex-specific growth charts [10]. Neonatal morbidities included intraventricular hemorrhage (IVH) grade 3–4 [11], necrotizing enterocolitis (NEC) requiring surgical intervention, retinopathy of prematurity (ROP) stage ≥3 [12], and severe bronchopulmonary dysplasia (BPD) defined as supplemental oxygen ≥30% or positive pressure [13].

Follow-up assessments at 18–22 months’ corrected age consisted of standardized neurological examination and administration of the Bayley Scales of Infant and Toddler Development, third edition (Bayley-III) by annually certified examiners [14]. Additional information was obtained through questionnaires, structured interviews, and review of children’s medical records. This information included: post-discharge hospitalizations, medical subspecialty care, medications, and surgeries; home medical equipment use; medical support service utilization; child care arrangements; and information on feeding. Data on hospitalizations, subspecialist visits, and surgeries reflected the duration between initial neonatal intensive care discharge and follow-up evaluation. Medication data reflected use during the 3 months prior to follow-up. Medication data were collected using categories defined by indication without specifying medication names or dosages. Anti-reflux medications included proton-pump inhibitors, H2 antagonists, and prokinetics. Medications for asthma or BPD included inhaled bronchodilators and corticosteroids. Anti-seizure medications included anticonvulsants; thyroid medications included levothyroxine; and anti-spasticity medications included baclofen. Information on medical equipment was recorded if the child used the equipment at follow-up or if it had been ordered for use at that time. Information on medical support service use at the time of follow-up, child care during the month prior to follow-up, and feeding habits at the time of follow-up were reported by the primary caretaker.

NDI classification

NDI was described using the definitions in Table 1. Components of NDI included the Bayley-III cognitive composite score, diagnosis of cerebral palsy, modified Gross Motor Function Classification System (GMFCS) score [15, 16], blindness and deafness. The Bayley-III cognitive composite score has a standardized mean of 100 with standard deviations (SDs) of 15 points, which have been used to define NDI cutoffs (“moderate NDI” being defined as >1 SD below the standardized mean [score <85], “severe NDI” as >2 SD below the standardized mean [score <70]) [17, 18]. Because of controversy about the appropriateness of normative reference cutoffs for Bayley-III scores in our population [19], a “potential/mild NDI” category was defined as a cognitive composite score ≤1 SD below the mean to avoid potential inclusion of children with cognitive composite scores near those used to define NDI in the “no NDI” group. All Bayley-III norms were based on the child’s adjusted age. Per NRN protocols, a modified GMFCS score was determined for all children and ranged from 0 (no abnormalities) to 5 (most severe), with the category of “possible level 1” being reserved for children with toe walking or asymmetric walking, suggesting the potential for mild diplegia or hemiplegia but not diagnostic of cerebral palsy at 18–22 months’ corrected age [20]. The latter was also used to characterize the “potential/mild NDI” group in order to avoid including children with non-diagnostic examiner concerns about motor function in the “no NDI” group. Functional blindness was defined by caretaker report with an exam consistent with bilateral visual acuity <20/200 despite correction. Functional deafness was defined as severe, bilateral hearing loss with or without amplification per caretaker report and consistent with the exam.

Table 1 Classification of Neurodevelopmental Impairment.
Full size table

Statistical analysis

Differences in the prevalence of outcomes among children in the four NDI groups (none, potential/mild, moderate, and severe NDI) were compared using Cochran-Mantel-Haenszel tests (for categorical variables) and ANCOVA (for continuous variables), as appropriate. The analyses were adjusted for gestational age at birth (in completed weeks), known to be correlated with NDI classification and expected to correlate with several outcomes. A p value <0.05 was considered statistically significant. Corrections for multiple comparisons were not conducted. All analyses were conducted using SAS version 9.4.

Results

There were 6542 children born at 22–26 weeks’ gestation who were eligible for inclusion. Of these, 2618 (40%) died and 441 (7%) had incomplete follow-up information. Of the remaining 3483 children, 825 (24%), 1576 (45%), 657 (19%), and 425 (12%) had no, potential/mild, moderate, and severe NDI, respectively.

Components of NDI among the children in this cohort are described in Table 2. In all categories, the most prevalent determinant of NDI classification was the Bayley-III cognitive composite score. Cerebral palsy was diagnosed in 6%, 18%, and 49% of children with potential/mild, moderate, and severe NDI, respectively. Blindness and deafness were present among 12% and 24% of children, respectively, with severe NDI.

Table 2 Rates of Each Component of Neurodevelopmental Impairment.
Full size table

Among infants surviving to follow-up, 17/27 (63%), 106/253 (42%), 293/786 (37%), 348/1122 (31%), and 318/1295 (25%) of children born at 22, 23, 24, 25, and 26 weeks’ gestation had moderate or severe NDI. As shown in Table 3, severity of NDI was associated with lower birth weight, male sex, 1-minute Apgar score ≤3, no exposure to antenatal corticosteroids, small for gestational age, maternal self-identification as non-white race, and lack of private insurance (p < 0.001). Severity of NDI was also associated with increased rates of grade 3–4 IVH, surgical NEC, stage ≥3 ROP, and severe BPD.

Table 3 Subject characteristics by neurodevelopmental impairment.
Full size table

NDI severity was associated with several concurrent outcomes at 18–22 months’ corrected age overall (Table 4) and at each gestational age (Supplemental Tables A1–A4). Rehospitalization following discharge from the neonatal intensive care unit was common for all children born extremely preterm, regardless of NDI severity, with 331/825 (40%) of children with no NDI and 300/425 (70%) of children with severe NDI rehospitalized by 18–22 months’ corrected age. Children without NDI were rehospitalized, on average, 1.7 times (SD 1.3) before 18–22 months’ corrected age and children with severe NDI 3.2 (SD 3.1) times. The most common reason for rehospitalization was respiratory disease. Regardless of NDI status, large proportions of children in each NDI category (17–31%) used medications at 18–22 months’ corrected age to treat asthma or BPD. Similarly, tympanostomy tube placement (rates of 9–16% across NDI categories) and hernia repair (rates of 11–17% across NDI categories) were common among all children born extremely preterm, regardless of NDI status.

Table 4 Childhood Outcomes at 18–22 Months’ Corrected Age by Neurodevelopmental Impairment.
Full size table

While nearly one quarter (82/361) of children with severe NDI had gastrostomy tube placement by 18–22 months’ corrected age, there was evidence of functional feeding difficulties in all NDI categories and 2% (15/725) of children with no NDI had a gastrostomy tube placed. With no NDI, 96% of children could feed themselves at 18–22 months’ corrected age, whereas 44% of children with severe NDI were able to feed themselves.

Use of medical equipment and medical support services was more common among children with severe NDI. Among children without NDI, 17% utilized physical and occupational therapy at 18–22 months and 21% utilized speech therapy services. These compare to 75% and 53%, respectively, among children with severe NDI. The utilization of traditional home-based or center-based daycare was much more common among children with no NDI (33%) than those with severe NDI (11%).

Discussion

In a cohort of children born extremely preterm, NDI severity was associated with increased rates of rehospitalization, medical subspecialty visits, surgeries, medication and equipment use, specialty medical child care, and functional feeding difficulties at 18–22 months’ corrected age. However, children born extremely preterm with no diagnosis of NDI also had high rates of rehospitalization, surgery, and medical utilization.

Childhood outcomes of extremely preterm infants have been extensively studied [21]. However, few others have described non-NDI outcomes in relation to NDI diagnosis [22, 23]. Our study raises important questions about how to best design and report studies of the outcomes of extremely premature birth. At a workshop held jointly by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), Society for Maternal–Fetal Medicine, American Academy of Pediatrics, and American College of Obstetricians and Gynecologists, participants recommended that “Physicians should recognize that the parents’ views on what is a ‘severe’ disability may be different from those of the researchers or clinicians…” [24] Others, including the Core Outcomes in Neonatology group, have worked with key stakeholders, including families, healthcare providers, and members of society, to evaluate which outcomes of prematurity are of importance to these stakeholders to collect and report in neonatal studies [25,26,27]. These outcomes include respiratory illnesses, readmissions, multiple operations, and the effects of a child’s illness on the family [26].

Strengths of this study include the large number of infants and high rates of follow-up in the NICHD Neonatal Research Network, as well as the availability of standardized data obtained by annually certified research personnel to diagnose NDI. The data included outcomes from academic medical centers across the United States representing various geographies and populations. NDI data from the NRN have been widely used in prognostic studies and clinical trials to help guide care of extremely preterm infants in the United States and around the world [28,29,30,31].

Limitations of this study include the post-hoc design of the analysis, which used only available data and precluded the ability to obtain additional information on family and parental quality of life, employment, education, relationships, health, and other outcomes. Moreover, this study was intended to be descriptive and was not designed to determine which factors affected both NDI and other outcomes at follow-up, such as neonatal morbidities or socioeconomic hardships. Importantly, follow-up data were limited to those collected at 18–22 months’ corrected age and provide only a snapshot from that time. NDI diagnoses may vary over children’s lives and assessments at later ages may have resulted in some children changing NDI categories [32, 33]. However, although 18–22 month outcomes may not adequately reflect long-term function or needs, data collected at 2 years’ corrected age are frequently used for perinatal decision-making [1].

Unlike many other neonatal morbidities, such as IVH [11] and retinopathy of prematurity [12], NDI does not have a consensus definition. Published definitions vary widely across studies [34]. Small variations in the definition of NDI can have a substantial influence on its rate in a population and on its association with specific variables [35, 36]. Despite this, studies of NDI in children born extremely preterm are used as the basis for recommendations to make treatment decisions, including whether to direct care toward survival or palliation [1, 24], and are frequently used as a component of the primary outcome of major clinical trials [2]. For the purposes of our study, we presented results separately for “no NDI” and “potential/mild NDI.” While the clinical significance of the “potential/mild NDI” category is debatable, at least one group has suggested using cutoffs as high as 95 for the Bayley-III cognitive composite score to indicate problems with cognitive development [37]. In our study, the designation was used to avoid misclassification of marginal cases in the “no NDI” group, which was important to study adverse outcomes among children without NDI. The definitions of NDI used in this study are similar to those used elsewhere [15] but cannot be compared with studies using different NDI definitions or criteria.

In conclusion, we found that severity of NDI was associated with several other important medical, functional, and social outcomes at 18–22 months’ corrected age. However, children born extremely preterm without an NDI diagnosis had substantial medical needs following discharge that may significantly impact families and the healthcare system. These data should be useful to support counseling families and the design of studies for early childhood outcomes beyond NDI.