Objective. Because the survival rate has increased for extremely low birth weight neonates, many have raised the concern that the rate of developmental disability among survivors will also increase. To address this concern, we analyzed changes over time in survival and major neurosensory impairment in a sample of extremely low birth weight infants born between July 1, 1979, and June 30, 1994.
Methods. The study sample included 513 infants with birth weights of 501 to 800 g who were cared for in either of the two neonatal intensive care units that serve a 17-county region in northwest North Carolina and who were born to mothers residing in that region. At 1 year of age (corrected for gestation), survivors were examined by a pediatrician and were tested using the Bayley Scales of Infant Development. Major neurosensory impairment was defined as cerebral palsy, a Bayley Mental Developmental Index <68, or blindness. A total of 209/216 (97%) of survivors were examined at 1 year of age. Epoch of birth was defined as follows: epoch 1, July 1, 1979 to June 30, 1984; epoch 2, July 1, 1984 to June 30, 1989; and epoch 3, July 1, 1989 to June 30, 1994.
Results. Survival rates for epochs 1, 2, and 3 were, respectively, 24/120 (20%), 63/175 (36%), and 129/218 (59%). In contrast, the proportions with a major neurosensory impairment did not increase over time; rates for successive epochs were 6/24 (25%), 17/61 (28%), and 26/124 (21%). Rates of cerebral palsy were 3/24 (13%), 12/61 (20%), and 9/124 (7%); rates of delayed mental development were 4/24 (17%), 12/61 (20%), and 17/124 (14%); and rates of blindness were 2/24 (8%), 0/62, and 5/124 (4%), respectively.
Conclusions. This analysis suggests that the increasing survival of extremely low birth weight neonates since the late 1970s has not resulted in an increased rate of major developmental problems identifiable at 1 year of age.
Numerous studies have indicated that during the 1980s, the survival probability increased for extremely low birth weight (ELBW) neonates.1-5 The relative contribution to the increasing survival of antenatal steroids, a more aggressive standardized approach to delivery room resuscitation of extremely premature infants, surfactant replacement, dexamethasone treatment for ventilator dependency, and other interventions is unknown. The risk of major neurosensory impairments among surviving ELBW infants has been found to be stable1-4 or decreasing across time.5 Despite these reassuring reports, concern remains that as the survival rate of ELBW continues to increase, rates of morbidity also will increase.6-9 In this study, we address this concern by examining secular trends in survival and major developmental disability during the 15 years between July 1, 1979, and June 30, 1994.
Between July 1, 1979, and June 30, 1994, 513 infants with birth weights of 501 to 800 g were born to mothers residing in a 17-county region of northwest North Carolina and were admitted to one of the two neonatal intensive care units serving that region. By matching data on these infants with birth certificate data for all North Carolina births in the years 1984 to 1987 (provided by the North Carolina Department of Health Statistics), we found that the study sample includes >90% of all surviving infants with birth weights of 501 to 800 g, born January 1, 1984, to December 31, 1987, to mothers residing in the 17-county region from which our sample was drawn.
All analyses of survival were based on the sample comprising all live births at our level III perinatal referral center, plus all infants transported and admitted to either of our two level III neonatal intensive care units. Almost 90% of the study infants were born in a level III perinatal referral center, which is also the site of one of the two neonatal intensive care units where patients were hospitalized. The birth of these infants, whom we refer to as inborn infants, was attended by a neonatologist, except when delivery was precipitous. Infants with birth weights >500 g who showed signs of life typically were resuscitated. Exceptions to this policy of aggressive resuscitation included infants known to have major congenital malformations or infants thought to be less mature than the perceived limit of viability. In the first part of the 1980s, this limit was thought to be 25 weeks; in the latter part of the 1980s, 24 weeks; and during the 1990s, 23 weeks.
Demographic data for neonates were abstracted from the medical record (eg, birth weight, gestational age [GA]). Before 1990, assignment of GA by the attending neonatologist was based on the obstetrician's estimate or on an assessment of the neonate.10 After 1990, the assignment was based on the date of the mother's last menstrual period unless this was not available, in which case, an obstetrician's estimate was used. In a study by Sanders et al,11 agreement between GA based on last menstrual period and that based on the best obstetric estimate was high, with exact agreement in 82% of cases. (In that study, best obstetric estimate was based on last menstrual period if this agreed within 2 weeks with GA based on fetal ultrasonography; otherwise, the estimate was based on ultrasonography.) When no prenatal estimate was available, GA was based on an assessment of the neonate. Data reported by Lubchenco et al12 were used to classify infants as small for GA. Data about neonatal treatments (eg, surfactant, dexamethasone) and medical complications (eg, pulmonary interstitial emphysema, intraventricular hemorrhage) were collected on forms completed by the attending neonatologist at the time of discharge from the hospital.
Chronic lung disease was defined as the use of supplemental oxygen at 36 weeks' postconception.13 Based on work by Stewart et al,14 we defined a major cranial ultrasound abnormality as either 1) complicated periventricular hemorrhage (subependymal/intraventricular hemorrhage associated with posthydrocephalus requiring placement of a shunt or persistent but nonprogressive ventricular dilatation; or 2) intraparenchymal echo density (echo density extending from the germinal layer or ventricles into brain tissue [consistent with parenchymal extension of periventricular hemorrhage, periventricular hemorrhagic infarction, or periventricular leukomalacia] or echo density superior and lateral to the ventricles in the periventricular white matter.15 The reliability of interpretations of cranial ultrasound examinations at our medical center has been reported elsewhere.16
All survivors were enrolled in an Infant Follow-up Project, which was begun in 1976 and has operated continuously since that time, with funding from the North Carolina Department of Environment, Health, and Natural Resources. Parents or care givers for infants in the project were contacted by letter and/or telephone in the first month after discharge and at 6 and 9 months adjusted age. At 12 months adjusted age, infants were scheduled for a multidisciplinary evaluation that consists of a hearing screen, administration of the Bayley Scales of Infant Development Scale,17 and the Vineland Adaptive Behavior Scale18 by clinical child psychologists, and a physical and neurologic examination by a developmental pediatrician, a neonatologist with special interest in developmental follow-up, or a pediatric resident supervised by a developmental pediatrician. After October 1993, the Bayley Scales of Infant Development Scale, 2nd ed, was used.19 The pediatricians were aware of infants' medical history at the time of examination. Psychologists were aware only of the infants' GA at birth, and after 1990, they obtained this information only after completion of developmental testing. As a general rule, nonambulating infants, infants suspected of having a neurologic abnormality, and infants with delayed motor development (based on the Bayley Scales Psychomotor Developmental Index value of <68) were seen by the developmental pediatrician or the neonatologist and a pediatric physical therapist. Cerebral palsy (CP) was diagnosed only if both a pediatrician and a physical therapist agreed on the presence of definitely abnormal position and posture attributable to impaired neuromotor function. Three physical therapists were involved in confirmation of diagnoses of CP. All three are pediatric physical therapists who staff a pediatric rehabilitation hospital. With a few exceptions, subjects with CP had the spastic form of this disorder. In general agreement with others,1,4,21 we defined a neurosensory impairment as a Bayley Mental Developmental Index (MDI) value <68 (ie, 2 SD units below the mean), CP, blindness, or deafness. At 1 year of adjusted age, infants were classified as having weight, length, or head circumference below the 10th percentile using gender-specific growth curves adapted from Hamill et al.20
Trends across time were assessed by comparing rates for groups of infants after categorizing them according to date of birth, as follows: epoch 1, July 1, 1979 to June 30, 1984; epoch 2, July 1, 1984 to June 30, 1989; epoch 3, July 1, 1989 to June 30, 1994. The χ2test for linear trend was used with a P < .05 used to define statistical significance. Associations between infant attributes and categorical outcomes were expressed as odds ratios (OR) and 95% confidence intervals (CIs) derived from logistic regression. Logistic regression was also used to evaluate the effect of date of birth, modeled as a continuous variable, adjusting for infant attributes. The assumptions of logistic regression (linearity of the effect of continuous variates on the log odds and additivity of the effects of covariates) were examined as described by Harrell and Lee.22 The selection of variables for inclusion in logistic regression models was based on the following two criteria recommended by Greenland23: 1) factors for which inclusion in regression models resulted in at least a 10% change in the estimated OR, and 2) factors that have been observed previously to have moderate or strong effects on neonatal mortality or the risk of developmental problems in ELBW infants.
As shown in Table 1, the distribution of birth weight, GA, gender, and race did not change during the 15-year study. Survival rates increased over time, from 20% in epoch 1 (24/120) to 36% (63/175) in epoch 2 and 59% (129/218) in epoch 3 (P < .0001, test for linear trend). The GA of survivors was significantly higher in the first epoch, and the proportion of married mothers was significantly lower in the third epoch.
Among all 513 study infants, the survival rate was higher for females and increased with increasing birth weight and GA. In multivariate analysis, survival was associated independently with these three infant characteristics, as well as with surfactant receipt (multivariate OR, 0.46 [0.23, 0.92]), dexamethasone treatment (multivariate OR, 0.30 [0.13, 0.71]), and pulmonary interstitial emphysema (multivariate OR, 6.06 [3.21, 11.43]). Even when adjustment is made for all of the factors listed, the likelihood of survival increased over time during the 15-year study (multivariate OR per year, 0.86 [0.80, 0.93]) (Table 4).
In terms of the neonatal complications listed in Table2, surviving infants were similar across the three epochs, with the exception of a lower proportion with chronic lung disease among infants born between 1979 and 1984. Of the 216 study infants who survived to 1 year of age, 209 (97%) of survivors were examined at 1 year of age. The rate of CP was lower in epoch 3 (P = .03, compared with epochs 1 and 2), whereas rates of Bayley MDI value <68 and the rates of any major neurosensory impairment (defined as CP, Bayley MDI value <68, or blindness) did not change over time (Table 3).
Date of birth was not associated with the risk of major neurosensory impairment. Only three of the factors listed in Tables 1 and 2 were associated independently with major neurosensory impairment. Adjusting for GA, which was inversely related to risk, major cranial ultrasound abnormalities were associated with an increased risk (multivariate OR, 5.71 [2.20, 14.84]), whereas years of maternal education was associated with a decreased risk (multivariate OR per additional year of education, 0.82 [0.67, 1.0]) (Table4).
GA-specific risks of adverse outcomes for infants born in epoch 3 are shown in Table 5. Among this group of infants, the association of GA with a major neurosensory impairment is attributable to the high risk among infants born at 24 weeks' GA.
The most important finding of this study is that as the survival rate of ELBW neonates increased from 20% to 59% between 1979 and 1994, no increase was found in the rate of major developmental problems identifiable at 1 year of age. Other studies of ELBW infants1,3-5 have also reported an increasing rate of survival during the 1980s. In agreement with Hack et al,2the present study indicates that this trend has continued during the 1990s. Our finding that increasing survival rates have not been accompanied by increasing rates of developmental problems also agrees with previous studies.1-5
Our study provides only limited information about the causes of the increasing survival for ELBW neonates. In multivariate analysis, surfactant replacement and prolonged treatment with dexamethasone were associated with decreased mortality. In contrast to the results of clinical trials of surfactant in infants <750 g,24,25 a large observational study indicated that a reduction in mortality occurred in this group after the introduction of surfactant.26 Clinical trials of dexamethasone typically have had limited statistical power to detect an effect on mortality, either because of small sample size27-29 or crossing over from placebo treatment to dexamethasone.30 The finding of a secular trend in mortality even after adjustment for infant characteristics (GA, gender, and birth weight) and the two therapies just noted suggests an influence from factors not considered in our analysis, such as increasing use of antenatal steroids or a decrease over time in the perceived limit of viability (with respect to GA).
A limitation of our study is that developmental outcome was evaluated at 1 year of adjusted age, before the age when some developmental problems, such as mild mental retardation and learning disabilities, can be diagnosed. Nonetheless, a recent report by Roth et al21 suggests that in ELBW infants, there is a high degree of concordance between the presence or absence of a major developmental impairment at 1 year of age and at 8 years of age. In that study, only 1 of 164 children who were categorized as having no impairment at 1 year of age was found to have a major impairment at 8 years, and a major impairment persisted in 21 of 23 children who were categorized this way at 1 year of age.21 Thus, we regard the absence of a secular trend in developmental problems at 1 year as evidence against there being a trend in problems manifesting later in childhood.
During the study, the Bayley Scales of Infant Development17were revised, so that approximately half of infants born in the third epoch (July 1, 1989 to June 30, 1994) were tested with the Bayley Scales of Infant Development, 2nd ed.19 Our research31 and the test manual19 indicate that infant scores on the Bayley Scales, 2nd ed, are lower than scores from the initial version of the test. Thus, any bias arising from the use of the Bayley Scales, 2nd ed, during the third epoch would tend to favor the two earlier epochs.
The clinical relevance of our findings pertains to the question of whether aggressive resuscitation of ELBW infants in the delivery room should be withheld from certain groups of infants. Because the survival rate for ELBW infants is now >50%, withholding care seldom can be justified on the assumption that such resuscitation ultimately will be futile. Most deaths in ELBW infants occur during the first 3 days of life;32 thus, withholding care can seldom be based on the concern that care for ELBW infants who die is unusually expensive. Finally, because we have found that within the group of infants weighing 500 to 800 g, birth weight is not a significant predictor of neurodevelopmental impairment; withholding resuscitation from the smallest ELBW infants cannot be justified on the supposition that these infants are at inordinately high risk for developmental disability. On the other hand, we found, as did Allen et al, that GA was associated with both survival and developmental outcome.6 Based on cranial ultrasound findings, Allen et al estimated that only 2% of their patients born at 23 weeks survived without developmental problems.6 In contrast, in our sample of infants born from 1989 to 1994, 27% (7/26) of infants with a GA of 23 weeks were alive and free from major neurosensory impairment. However, this outcome was found for only 8% of infants with a GA of 22 weeks. Because we rarely provide aggressive efforts at resuscitation for infants born at 22 weeks' gestation, we cannot speculate as to how much improvement in outcome is possible with such infants.
Recently reported research has described interventions that may reduce the rate of developmental problems among infants at the limit of viability. In agreement with others,33,34 we found that the risk of major neurodevelopmental impairment is associated with severe cranial ultrasound abnormalities and with maternal education, an indicator of the nurturance in the home environment. Thus, interventions to prevent intraventricular hemorrhage, such as postnatal indomethacin35 and prenatal phenobarbital,36and interventions to improve the infant's developmental environment37,38 could lead to additional improvement in the outcome of ELBW infants.
This work was supported in part by the North Carolina Department of Environment, Health, and Natural Resources, and Brenner Children's Hospital (Winston-Salem, NC).
- Received February 5, 1997.
- Accepted May 29, 1997.
Reprint requests to (T.M.O) Department of Pediatrics, Bowman Gray School of Medicine of Wake Forest University, Medical Center Blvd, Winston-Salem, NC 27157.
- ELBW =
- extremely low birth weight •
- GA =
- gestational age (in weeks) •
- CP =
- cerebral palsy •
- MDI =
- Bayley Mental Development Index •
- OR =
- odds ratio •
- CI =
- confidence interval
- Hack M,
- Friedman H,
- Fanaroff AA
- Robertson CMT,
- Hrynchyshyn GJ,
- Etches PC,
- Pain KS
- La Pine TR, Jackson C, Bennett FC
- Silverman W
- ↵Tyson J. Evidence-based ethics and the care of premature infants. In: Behrman RE, ed. Low Birth Weight. Los Altos, CA: Center for the Future of Children, David and Lucile Packard Foundation; 1995
- Sanders M,
- Allen M,
- Alexander GR,
- et al.
- Lubchenco LO,
- Hansman C,
- Boyd E
- Shennan AT,
- Dunn MS,
- Ohlsson A,
- Lennox K,
- Hoskins EM
- ↵Bayley N. The Bayley Scales of Infant Development. New York, NY: Psychological Corporation; 1969
- ↵Sparrow SS, Balla DA, Cicchetti DV. Vineland Adaptive Behavior Scales. Circle Pines, MN: American Guidance Service, Inc; 1984
- ↵Bayley N. Bayley Scales of Infant Development. 2nd ed. San Antonio, TX: Psychological Corporation; 1993
- Hamill PVV,
- Drizd Ta, Johnson CL, et al
- ↵Harrell FE, Lee KL. The practical value of logistic regression. In: Proceedings of the Tenth Annual SAS Users Group International Conference; Cary, NC: SAS Institute; 1985:1031–1036
- Smyth J,
- Allen A,
- MacMurray B,
- et al.
- Horbar JD,
- Wright EC,
- Onstad L,
- and the Members of the National Institute of Child Health and Human Development Neonatal Research Network
- Rastogi A,
- Akintorin SM,
- Bez ML,
- Morales P,
- Pildes RS
- Durand M,
- Sardesai S,
- McEvoy C
- Goldstein DJ,
- Fogle EE,
- Wieber JL,
- O'Shea TM
- Meadows W,
- Reimshisel T,
- Lantos J
- Ment LR,
- Oh W,
- Ehrenkranz RA,
- et al.
- Shankaran S,
- Cepeda E,
- Muran G,
- et al.
- Naglieri JA
- Copyright © 1997 American Academy of Pediatrics