Neurodevelopmental Outcomes of Extremely Low Birth Weight Infants <32 Weeks’ Gestation Between 1993 and 1998

METHODS

Perinatal data were collected prospectively by study nurses using standard registry forms.^4,5 Families were invited to participate in their center follow-up programs for a comprehensive assessment that consisted of a battery of developmental, neurologic, and behavioral assessments; a medical and social history; and parent interviews at 18 to 22 months’ corrected age. All definitions, protocols, and procedures are contained in a manual of operations at each study center and have been reported.^4,5 The network used predefined criteria for complete follow-up for this study. A successful visit was defined as medical history, interviews, anthropometrics, and either the Bayley Scales of Infant Development II (BSID-II)⁶ or a neurologic examination and gross motor assessment. Details of the procedures have been published.^4,5 The BSID-II⁶ was administered by a certified examiner who was trained to reliability by 1 of 4 psychologists with high interexaminer reliability and previous formal training in test administration. The Mental Developmental Index (MDI) and the Psychomotor Developmental Index (PDI) were derived. Certification by centers for administration of the BSID-II was achieved by the successful completion of 2 videotaped demonstrations of accurate performance and scoring of the BSID-II as determined by 1 of the 4 Gold Standard psychologists. Certification is renewed annually for all testers.

The neurologic examinations were based on the Amiel Tison⁷ neurologic assessment. Neurologic examinations were performed by experienced, certified examiners who had been trained to reliability in a 2-day workshop on the neurologic assessment. When a new neurologic examiner was added at a center, the certified examiner at the center trained the new staff member. Cerebral palsy (CP) was defined as a nonprogressive central nervous system disorder characterized by abnormal muscle tone in at least 1 extremity and abnormal control of movement or posture. Moderate to severe CP included children who were nonambulatory or required an assistive device for ambulation. Structured interviews were performed to obtain socioeconomic status information,⁸ including maternal and paternal education and occupation, marital status, and a detailed interim medical history including data on hearing status and vision status.

RESULTS

Table 1 shows the survival and follow-up rates for the 2 gestational age groups over the 3 epochs. Survival to hospital discharge improved significantly over the 3 epochs, from 55% to 61% for infants who were born at 22 to 26 weeks’ gestation and from 82% to 86% for infants who were born at 27 to 32 weeks’ gestation. Mean birth weight did not change over time. Survival rates, as expected, were higher for the 27- to 32-week gestational age group. The percentage of survivors who received a complete assessment on the basis of follow-up study criteria improved over time. Although infants who were lost to follow-up had some increased risk factors (3.0% more outborn, 3.0% less prenatal care, 10.2% less antenatal steroids, and 4.5% less surfactant use), they also had lower biological risk with higher birth wt (12.1 g), less chronic lung disease (4.3%), lower percentage of multiple birth (3.5%), fewer days in hospital (4.6 days), fewer postnatal steroids (9.8%), and fewer days on a ventilator (4.3 days). The subjects shown in Tables 2 through 4⇓⇓⇓ were survivors with a complete study follow-up visit.

Table 2 shows the changing demographic characteristics of the study sample evaluated at 18 to 22 months over the 3 study epochs. Although the percentage of teenage mothers remained relatively stable, the rates of low maternal education (<12 years), Medicaid insurance, outborn status, grades 3 to 4 IVH, and days in the hospital decreased significantly. In addition, the percentage of white mothers, those who received prenatal care, the cesarean section rate, surfactant administration, oxygen at 36 weeks, and multiple births increased significantly over time in both gestational age groups. The percentage of multiple births increased from 18.3% to 24.0% and from 20.9% to 25.6% for the early- and late-gestation groups, respectively (P < .0001). Significant gestational age effects were seen for all of the neonatal characteristics except for outborn and multiple birth. Those in the 22- to 26-week gestation group were at greater environmental risk with more mothers with <12 years of education, more mothers on Medicaid, and a lower rate of prenatal care. They also consistently had higher rates of neonatal morbidities. Analyses not shown indicated that 85% of multiples were twins, and 15% were higher order multiples. Additional analyses were run to evaluate the socioeconomic status characteristics of mothers of multiple births compared with singletons. Mothers of multiples were more likely to have an education ≥12 years (P < .0421), have private insurance (P < .0001), have age ≥19 years (P < .0001), and be married (P < .0001).

The 4 interventions of interest that were administered to the survivors who were seen at 18 months are shown in Table 3. The rates of administration of antenatal steroids and the use of high-frequency ventilation increased significantly over the 3 epochs. Postnatal steroid use increased during epoch 2 and decreased during epoch 3 for both gestational age groups. The number of days to regain birth weight decreased from 19.1 days to 17.5 days for the early-gestation group and from 15.1 days to 13.2 days for the late-gestation group (P < .0001). Significant gestation effects were identified. The 22- to 26-week gestation group had lower rates of antenatal steroid administration, higher rates of postnatal steroid administration and high-frequency ventilation, and longer duration to regain birth weight.

Table 4 shows the neurodevelopmental outcomes at 18 to 22 months of age. Rates of any CP and moderate to severe CP did not change significantly over the 3 epochs. Rates of other severe sequelae, including blindness, PDI <70, MDI <70, and NDI, decreased significantly across the 3 epochs. The percentage of children who required amplification for hearing loss or shunt for hydrocephalus or had seizures at 18 to 22 months remained unchanged. It was not unexpected that rates of adverse neurodevelopmental morbidities were higher in the 22- to 26-week gestation group. Only seizures did not differ among study gestation groups.

Figure 1 shows the logistic regression analysis to predict moderate to severe CP. Variables that were associated with an increased risk were PVL, grades 3 to 4 IVH, BPD, and male gender. In addition, 1 center had an increased risk for CP when compared with the reference center. After confounders were adjusted for, antenatal steroids were associated with decreased risk (OR: 0.66; 95% CI: 0.47–0.92) and postnatal steroids with increased risk for moderate to severe CP (OR: 2.02; 95% CI: 1.40–2.92).

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Logistic regression for moderate to severe CP, ORs and 95% CIs. Logistic regression models to predict moderate to severe CP, PDI <70, MDI <70, and NDI were completed. Variables that were entered into the regression analyses included epoch; gestational age group; birth weight; gender; SGA versus appropriate for gestational age; multiple births; surfactant; grades 3 to 4 IVH; PVL; sepsis; oxygen requirement at 36 weeks; white versus nonwhite race; outborn versus inborn status; cesarean section versus vaginal delivery; maternal education <12 years versus ≥12 years; private health insurance versus public; conventional ventilation versus none; adjusted age at the time of assessment; center; and the 4 interventions of interest: antenatal steroids (yes, no), high-frequency ventilation versus none, days to regain birth weight, and postnatal steroids (yes, no).

Figure 2 shows the logistic regression analysis to predict PDI <70. Birth weight and higher adjusted age at the time of the assessment were associated with decreased risk. Factors that were associated with increased risk included PVL, BPD, grades 3 to 4 IVH, multiple birth, maternal education <12 years, male gender, sepsis, and center. High-frequency ventilation just missed significance for increased risk (OR: 1.294; 95% CI: 0.990–1.691; P < .0589). Eight centers had an increased risk compared with the reference center. After adjustment for all other variables, antenatal steroids remained associated with decreased risk (OR: 0.66; 95% CI: 0.52–0.84) and postnatal steroids with increased risk (OR: 1.99; 95% CI: 1.56–2.55).

• Download figure
• Open in new tab
• Download powerpoint

Fig 2.

Logistic regression for PDI <70, ORs and 95% CIs.

Figure 3 shows the logistic regression to predict MDI. Higher birth weight was associated with decreased risk. PVL, BPD, multiple births, male gender, nonwhite race, maternal education <12 years, outborn, grades 3 to 4 IVH, center, and Medicaid insurance were associated with an increased risk. Three centers were at increased risk compared with the reference center. After adjustment for these variables, postnatal steroids (OR: 1.29; 95% CI: 1.04–1.61) and high-frequency ventilation were associated with increased risk. Epoch 2 versus epoch 1 was significantly associated with a decreased risk for MDI <70.

• Download figure
• Open in new tab
• Download powerpoint

Fig 3.

Logistic regression for MDI <70, ORs and 95% CIs.

Figure 4 shows the logistic regression to predict the presence of NDI. Higher birth weight was protective. Factors that were associated with increased risk included PVL, BPD, multiple birth, male gender, nonwhite race, grades 3 to 4 IVH, maternal education <12 years, Medicaid insurance, and center. Six centers were at increased risk for major impairment compared with the reference center. After adjustment for confounders, epoch 2 versus epoch 1 was associated with decreased risk; high-frequency ventilation and postnatal steroids were associated with increased risk.

• Download figure
• Open in new tab
• Download powerpoint

Fig 4.

Logistic regression for NDI, ORs and 95% CIs.

DISCUSSION

This study supports the importance of vigilant monitoring of both perinatal interventions and neurodevelopmental outcomes of ELBW infants over time. Multicenter data sets with large sample sizes provide a vehicle for identifying changes over time that would not be readily available with single-center analysis. Although rates of CP, hearing loss, hydrocephalus with shunt, and seizure disorders remained unchanged, the rates of low BSID-II scores, blindness, and overall NDI improved.

Four current clinically relevant interventions were identified for analysis for impact on outcomes: antenatal steroids, postnatal steroids, high-frequency ventilation, and number of days to regain birth weight. Significant evidence has accumulated in clinical trials of the beneficial effects of antenatal steroid administration,^9–12 and pivotal evidence supporting the administration of antenatal steroids was reported in 1995 by the National Institutes of Health Consensus Conference.¹² It is apparent that the evidence of beneficial effects has had an impact on the NICHD network centers, because the rate of administration of antenatal steroids increased significantly between epoch 1 (1993–1994) and epoch 2 (1995–1996). Multivariate analysis confirmed the independent association of increased antenatal steroid administration with lower rates of severe CP and low PDI.

Administration of postnatal steroids^13,14 continues to undergo scrutiny as a result of reports of the association with an increased risk for CP. The reports of Yeh et al^15,16 first established a link between postnatal steroid administration and adverse motor outcomes. In this report, postnatal steroid administration at sites was given for treatment of BPD and to achieve weaning from ventilators. Our analysis revealed an interesting time effect for both gestational age study groups. Although there was increased use of postnatal steroids between 1993–1994 and 1995–1996, there was a decrease in use in 1997–1998, suggesting a response to the studies of Yeh and others. In addition, adverse effects on neuromotor and cognitive function at school age have now been reported.¹⁷ Our data are consistent with previous studies demonstrating an association of increased postnatal steroid use with an increased risk for moderate to severe CP, low MDI, low PDI, and NDI. A limitation of our data on both antenatal and postnatal steroid administration is that we do not have specific information in our database on doses, number of courses of administration, and timing of administration.

Studies to date have not confirmed that the introduction of new types of ventilation techniques is better than conventional ventilators^18,19 for the prevention of BPD or NDI. The NICHD network data from 1993 to 1998 indicate a steady increase in the use of high-frequency ventilation from 12.3% to 33.1% for the early-gestation group and from 4.7% to 15.0% for the late-gestation group. Multivariate analyses revealed significant independent associations between high-frequency ventilation and NDI and MDI <70. Although this association may be related to the more common use of high-frequency ventilation for sicker, smaller infants with greater respiratory compromise, the finding is noted and deserves additional investigation. The use of high-frequency ventilation almost tripled during the 3 study epochs. Information on whether high-frequency ventilation use was for prevention or rescue was not available.

The final intervention that was evaluated was the number of days to regain birth weight, which was considered a proxy for early provision of nutrition. Although this is a weak marker for nutrition and subject to differences in fluid management, it has been shown to reflect nutritional management.^20,21 Nutritional intake was not available for the infants in this report. There has been considerable concern about the ability of current NICU protocols to provide adequate nutritional intake for ELBW infants in the first weeks of life.^20,21 The NICHD network reported that 95% of ELBW infants from the 1993–1994 era are below the 10th percentile for weight at 36 weeks’ postnatal age, reflecting postnatal growth restriction.^22,23 Efforts are being undertaken at centers to optimize nutritional intake. We did identify a clinically small but statistically significant decrease in the number of days to regain birth weight of 1.6 days for the early-gestation group and 1.9 days for the late-gestation group (P < .0001) during the 3 epochs. Days to regain birth weight was not associated with our outcomes in multivariate analyses. Prospective evaluation of the effects of actual nutritional intake on NDI is needed. In addition, although surfactant²⁴ and cesarean section both increased over the 3 epochs, neither was associated with 18–22 month study outcomes.

We note that the follow-up rate improved over the 3 epochs. We cannot rule out the possibility that a higher percentage of families with children whose outcomes were normal returned in recent epochs, thus contributing to our findings.

One of the neonatal variables included in our multivariate analyses was multiple birth. The rate of multiples has been increasing both in the United States and internationally.²⁵ The primary contributor to this increase is assisted reproductive technology (ART).²⁶ Our data show an increase of multiples from 18.3% to 24.0% for early-gestation infants and 20.9% to 25.6% for late-gestation infants (P < .0001), indicating that one quarter of ELBW infants <32 weeks’ gestation at network centers are multiples. Although previous studies^27–29 have not demonstrated a higher incidence of neurodevelopmental morbidities for multiples matched with control subjects, our logistic regression analysis indicated that multiples were a risk factor for low PDI, low MDI, and NDI among ELBW infants. A weakness of the NICHD data set is that ART/non-ART fertility methods were recently added to the data collection and were not available for this analysis. In the year 2000 in the United States, 82% of triplets and 33% of twins were conceived by ART/non-ART fertility methods.³⁰ We assume that the rates of ART procedures are reasonably similar within our observational cohort, and this finding deserves additional investigation.

We noted unexpected secular trends in the demographics of the study population. The percentage of mothers of white race increased, and the percentage of mothers with less than a 12th-grade education and the percentage with Medicaid insurance decreased.²⁵ The findings of decreased patients on Medicaid insurance may be related to federal government efforts to transition families off Medicaid and onto Medicaid health maintenance organization or private insurance.³¹ The percentage of outborn births also decreased significantly in both the early-gestation (13.1% to 9.1%) and late-gestation (11.7% to 8.6%) groups. Although outborn birth has previously been shown to be associated with lower rates of survival,^32,33 the only primary outcome with which outborn birth was associated was a higher rate of low MDI.^29,31 Socioeconomic variables were associated with our improved outcomes. The increased incidence of ELBW infants who were born to white, non-Hispanic, nonblack mothers may be related to the increased number of multiples over the 3 epochs and be a proxy for ART procedures. Nonwhite race was associated with increased rates of low MDI and NDI. Our data did confirm the increased vulnerability of the most immature (22–26 weeks’ gestation) ELBW infants. These infants experienced increased environmental risk and had higher rates of neonatal interventions, neonatal morbidities, and neurodevelopmental sequelae at 18 to 22 months compared with the 27- to 32-week gestation infants. Because the data set consists of infants with a birth weight ≤1000 g and a gestation <32 weeks, the larger gestational age cohort has a high percentage of SGA infants, which may also have an impact on our findings. Finally, center differences remained in all of our regression models after adjusting for demographics and interventions. This suggests that differences in center neonatal practices influence outcomes and is consistent with our previous report.⁵

It is apparent from our regression analyses that multiple factors play a role in determining outcome at 18 to 22 months. The interaction of demographic variables, neonatal variables, and interventions remains complex and undoubtedly will continue to change over time. The overall findings of our cohort at 18 to 22 months may be viewed with cautious optimism. Neurodevelopmental outcomes for ELBW infants who were born between 1993 and 1998 remained stable or improved. The changes in outcome at 18 to 22 months were associated with concurrent changes in demographics, interventions, and neonatal morbidities over the 3 epochs. Antenatal steroids was the only study intervention that was significantly associated with better outcomes at 18 to 22 months. Efforts to continue to improve outcome must include clinical trials to evaluate both interventions currently in use and novel interventions.