Neurodevelopmental and Functional Outcomes of Extremely Low Birth Weight Infants in the National Institute of Child Health and Human Development Neonatal Research Network, 1993–1994

DISCUSSION

This is the first multicenter, prospective report of outcome at 18 to 22 months' corrected age in a large cohort of ELBW infants born in the 1990s. The large sample provides sufficient numbers to report outcome of the cohort by 100-g birth weight subgroups and to adjust for multiple risk factors for adverse outcome. Only the 401- to 500-g subgroup is too small to be representative. The data were collected at 12 centers across the United States, representing a spectrum of social and demographic characteristics. Fifty-one percent of the cohort had a normal neurodevelopmental and sensory assessment. The probability of being classified as abnormal neurologically increased steadily as birth weight decreased. Whereas 25% of infants with birth weights of 901 to 1000 g had abnormal neurologic findings, this increased to 43% of infants 401 to 500 g. A specific diagnosis of CP based on the strict study criteria was made in 17% of the infants (range: 15% for infants 801–1000 g to 29% for infants 401–500 g). This incidence of CP at 18 to 22 months' corrected age is somewhat higher than previous reports (range: 7%–14.8%) of ELBW infants.^16–21 The incidence is similar, however, to the Emsley et al²² report of infants born at 23 to 25 weeks during 2 periods (1984–1989 and 1990–1994). They found the rate of disability increased from 38% to 68% and blindness increased from 4% to 18%, whereas the CP rate remained relatively stable at 21% and 18%. Disability in their study, however, also included mild findings, such as myopia and strabismus. The NICHD cohort had a mean gestational age of 26 weeks and was born in 1993 and 1994; therefore, they are reasonably comparable in time and gestation to the second period of Emsley et al. The incidence of significant vision impairment (3% blind) and hearing impairment requiring aids (3%) is consistent with other recent reports of ELBW infants,^16,,21 although earlier reports often identified greater numbers of visually impaired infants.^6,,19 This low incidence of blind infants in our study may be attributed to aggressive ophthalmologic monitoring and management within the neonatal units in the 1990s.

Developmental test scores obtained with the new Bayley II¹⁴ resulted in lower scores than previously reported from these units with the Bayley I. The Bayley II, which is based on new standardization data, was published in 1992. The restandardization was initiated because of evidence of an upward drift²³ in the mean scores on the old Bayley I.²⁴ The mean scores of the Bayley II MDI and PDI are 12 and 15 points lower, respectively, than the corresponding Bayley I score: the Bayley II mean MDI of 76 is equivalent to a Bayley I MDI of 88 and the Bayley II PDI score of 78 becomes a 93 on Bayley I. This may be a partial explanation for the low test scores reported in this study relative to studies published before 1992, in which the developmental outcomes were overestimated by the Bayley I.

The scores reported in this study are representative of 18 month performance of ELBW infants born in the 1990s. In the current study, 37% of the cohort had a Bayley II MDI more than 2 SD below the mean and 29% had a Bayley II PDI more than 2 SD below the mean (significant cognitive or motor delay). If milder developmental delay (1 SD below the mean) is used, then 66% and 57% of the children are at potential risk of developmental or motor morbidities. As in the functional outcomes, there was a trend for infants of lower birth weights to have lower test scores. An exception was the small subgroup of 14 infants <500 g. Since the subgroup was small and 2 infants in the group were not tested, their mean score may not be representative of infants <500 g. Very little data have been reported on these tiny infants. Most previous studies have reported 20–26% of ELBW infants with Bayley scores <70. Most of these studies, however, report on a total population of infants <1000 g or <1250 g and not on subgroups.^6,,19,21 Ment et al²¹ reported a rate of 26% for infants <1250 gms. Piecuch et al²⁵reported 18-month Bayley outcomes of children born at 24, 25, and 26 weeks' gestation between 1990 and 1994. Forty-two percent of children were administered the Bayley I and 58% the Bayley II. Scores >2 SD below the mean were identified in 39% of 24-week children, 30% of 25-week children, and 11% of 26-week children. The results support our findings that the lower gestation ELBW children are at increased risk of significant cognitive deficit. The percentage of MDI scores >2 SD below the mean, however, remains higher in our study. Although the low scores at 18 to 22 months' corrected age in our study sample raise great concern, it is unclear whether some of these ELBW infants are still in a catch-up phase or whether they will be destined to long-term learning and performance challenges. Our multivariate analyses did indicate that older age at testing within the 18- to 22-month window was associated with a higher PDI, suggesting motor catch-up is in progress. Assessments at 3 to 7 years of age are required to address this question. Studies in the 1990s reported that ELBW infants followed to school age are at significantly increased risk of academic difficulties, with 45% to 50% receiving resource or special education support services.^26–33

Functional assessment of daily living skills, including lower extremity mobility (walking) and upper extremity function (pincer grasp and feeding abilities), were also assessed. Overall, 70% of ELBW toddlers were walking fluently, 86% had a pincer grasp, and 80% were feeding independently. In comparison, term children consistently achieve inferior and superior pincer grasp at 9 to 12 months of age, independent walking at 12 to 18 months of age, and independent feeding at 15 to 18 months of age. As with the Bayley, there were similar trends for lower birth weight infants to have less functional skill development. Among 14 infants <500 g, only 57% were walking fluently, 64% had a pincer grasp, and 67% were able to feed independently. Others have shown that functional skills may be delayed in preterm infants.³⁴ For the total cohort, another 13% of infants had walking skills that were rated as less than fluent. Many of these infants were expected to be walking more fluently within a few months of the study assessment. The data of Msall et al³⁵on functional assessments of VLBW infants at 4.5 years of age using the WeeFIM System data suggest that catch-up is still occurring for VLBW infants at 18 months' corrected age.

Although most of the predictors of poor outcomes in the current study were anticipated, the data are unique because they assess the relative impact of multiple risk and protective factors for ELBW infants cared for with current management strategies. Lower birth weight, grades 3 to 4 IVH/PVL,^20,,35,37 and CLD^38,,40 were all associated with increased morbidity among the ELBW survivors. In addition, steroid treatment for CLD was significantly associated with abnormal neurologic status, Bayley PDI <70, Bayley MDI <70, and no independent walking. Previous studies^40–47 of the use of postnatal dexamethasone for CLD have reported mixed findings. In the randomized, double-masked, controlled trial by Yeh et al,^40,,44 early steroid therapy was associated with increased mortality and a significantly higher incidence of neurodevelopmental dysfunction at 2 years' corrected age. In the current study, use of postnatal steroids was associated with prolonged oxygen therapy, suggesting those infants had more severe CLD. The association with poor outcome, however, persisted after adjusting for CLD. It is of interest that although postnatal steroid therapy was a risk factor for an abnormal neurologic examination, it was not an independent risk factor for CP, indicating other factors (ie, IVH/PVL and CLD) are more directly associated. Eight percent of children had other neurologic findings, such as hypotonia, which are known to be associated with CLD and can affect the neurologic examination and motor performance. The mechanism by which postnatal steroids affect neurodevelopment remains unclear. Animal models^46,,47 suggest a direct toxic effect; further investigation in the form of prospective clinical trials to test the relative benefits and risks of postnatal steroids are indicated. Surfactant was identified as a risk factor for CP even after adjusting for birth weight and multiple other risk factors. Because the administration of surfactant is based on clinical judgment, and surfactant is given to sicker infants, we speculate that surfactant administration may be a marker for illness severity in the first hours of life.

Another risk factor associated with both an abnormal neurological examination and a low Bayley PDI was NEC. Although the pathophysiology of NEC is not clearly understood, the histologic appearance of the disease includes coagulation necrosis suggesting that ischemic events play a role.⁴⁸ It has been suggested that inflammatory mediators associated with bacterial invasion lead to vasoconstriction and hypoxic ischemic events. The associations among NEC, illness severity,⁴⁹ infection, surgical intervention, and outcome require further investigation.

Male gender also significantly increased the risk for an MDI <70. Msall et al⁵⁰ have previously reported that male gender was a predictor of disability in infants <28 weeks' gestation. Significant protective factors included higher birth weight (MDI), white race (MDI), and higher maternal education (MDI). This is consistent with previous reports showing that social and environmental factors impact on outcome.^51–53 This study demonstrated that a higher level of maternal education had a positive association with Bayley MDI scores as early as 18 to 22 months' of age. The positive effects of white race are also likely secondary to the strong linkage between poverty and black race in the United States.

Because of the known positive effects on survival and pulmonary maturation, we had speculated that prenatal steroid therapy would be associated with better neurodevelopmental outcome. Although prenatal steroid therapy was initially associated with a positive benefit on walking and the presence of a pincer grasp, the adjusted ORs after all risk factors were entered into the model were not statistically significant. Only 38% of the cohort, however, received prenatal steroids. Further prospective evaluation of the long-term effects of prenatal steroids is needed. Finally, although maternal hypertension had previously been shown to have both positive and negative effects on outcome, we identified no clear association with our outcomes. A known maternal risk factor^54–56 for poor neurodevelopmental outcomes that was not investigated in this study was maternal alcohol and substance abuse. Although the maternal population in this report is at high social risk (65% Medicare and 28% less than a high school graduate), it is difficult and costly to obtain accurate information, and a decision was made a priori not to collect this data. We did, however, include maternal education and race in the logistic regression models. Another risk factor of interest is the 19% incidence of multiples (17% twins and 2% triplets or greater).⁵⁷ It is well-known that the incidence of multiples resulting from assisted reproductive technology is increasing^58,,59 and that increased short-term and long-term morbidities are associated with higher order multiples.^58,,59 The percentage of higher order multiples in the current study was relative low (2%) and may be secondary to current efforts to limit the number of embryos transferred.

The 49% incidence of abnormal neurodevelopmental and/or sensory findings at 18 to 22 months of age supports the need for referral of all ELBW infants for early intervention services. The children at greatest risk for significant delay (MDI or PDI <70) were those with abnormal neurologic examinations.

Although our cohort of 1151 ELBW survivors from 12 tertiary centers is the largest current cohort with comprehensive neurodevelopmental data at 18 months of age, we note that the dataset has limitations secondary to the short duration of follow-up and the 78% follow-up rate. Follow-up to school age, although expensive and fraught with the challenges of long-term tracking of families, would provide additional information on both utilization of special education resources and academic success. In addition, Tin et al⁵⁷ identified an increased disability rate among noncompliant families who do not return for follow-up clinic visits, suggesting that the actual morbidity rate may be even higher than our report suggests. However, Tin et al⁵⁷ included in his definition of disability all children who died before their second birthday, eg, disabled was defined as death or disability. Most reports in the US literature separate death and disability. Examination of the characteristics of families lost to follow-up in a US population, however, deserves further evaluation. Although we agree that to achieve close to 100% follow-up would eliminate any possible bias of outcome interpretation, this goal remains difficult to achieve.