Abstract
BACKGROUND. Neurodevelopmental impairment of extremely low birth weight infants increased in the 1990s. Modern therapeutic changes may have influenced more recent neonatal outcomes.
OBJECTIVE. We sought to compare neonatal therapies and outcomes among all extremely low birth weight infants born in 2000–2002 (period III) to 2 previous periods: 1982–1989 (period I) and 1990–1999 (period II).
METHODS. The population included 496 extremely low birth weight infants born at our perinatal center during period I, 749 during period II, and 233 during period III. Therapies, rates of death, and survival with and without impairment at 20 months' corrected age were compared.
RESULTS. Between periods I and II, survival increased from 49% to 68% as did neonatal morbidity. This resulted in increased survival without impairment but also increased survival with impairment. Changes in therapy during period III included an increase in antenatal steroid use and a decrease in postnatal steroid use, although the rate of chronic lung disease did not change. Sepsis decreased, as did severe intraventricular hemorrhage. On follow-up, the rate of cerebral palsy decreased from 13% to 5%, resulting in a decrease in neurodevelopmental impairment from 35% to 23%. As a result, during period III versus II, survival without impairment increased, whereas survival with impairment decreased.
CONCLUSION. Since 2000, neurodevelopmental impairment has decreased among extremely low birth weight infants. A variety of perinatal and neonatal factors were associated with the improved outcomes including increased antenatal steroid use and cesarean section delivery, as well as decreased sepsis, severe cranial ultrasound abnormalities, and postnatal steroid use despite no change in the rate of chronic lung disease.
The evolution of neonatal intensive care therapy has resulted in dramatic improvements in the survival of extremely low birth weight (ELBW) infants (<1000 g).1 Although initial reports of declining mortality showed no increase in the rates of neurodevelopmental impairment,2,3 other studies suggested an increase in neurodevelopmental problems.4–6 Ethical concerns have been expressed that aggressive resuscitative efforts with tiny infants may result in increased rates of disability among survivors and substantial resource use.7,8 Additional reports suggest adverse effects of modern therapies such as postnatal steroid use.9 Although frequently used to treat chronic lung disease during the 1990s, the use of postnatal steroids has declined dramatically based on recommendations of The American Academy of Pediatrics.10 Quality improvement endeavors based on the sharing of best practice methods to decrease sepsis rates have also been implemented.11
Recently, we showed worsened neurodevelopmental outcomes for ELBW infants born in the 1990s, a period characterized by frequent postnatal steroid use and before the initiation of major efforts to decrease neonatal sepsis.12 In light of the recent changes in care practices, we sought to compare the rates of survival, neonatal morbidity, and neurodevelopmental impairment at 20 months' corrected age (CA) among ELBW children who were born at our perinatal center during 3 eras: 1982–1989, marked by enhanced assisted ventilation; 1990–1999, associated with the introduction of surfactant and antenatal and postnatal steroid therapy; and 2000–2002, marked by routine antenatal steroid therapy with decreased postnatal corticosteroid use and sepsis prevention initiatives.
METHODS
The study population included all 1478 live-born infants without major congenital malformations and with birth weights between 500 and 999 g who were delivered at our perinatal center, MacDonald Hospital for Women, during 3 periods (ie, 1982–1989 [period I; n = 496], 1990–1999 [period II; n = 749], and 2000–2002 [period III; n = 233]). The study periods were selected on a priori grounds on the basis of the following factors: the availability of uniform cranial ultrasound data for all patients born from 1982 onward, the initiation of antenatal steroid and surfactant therapies after 1990, and the implementation of quality improvement measures aimed at reducing postnatal corticosteroids and late-onset sepsis in 2000. Complete neurodevelopmental assessments to 20 months' CA were available for children born through 2002.
Our perinatal center is a tertiary referral center for high-risk pregnancies. Neonatal specialists attend all preterm deliveries. Decisions concerning active treatment of extremely preterm infants are made after consultation with the family, if possible, and/or according to the condition of the infant at birth. Despite some minor variations among individual neonatologists, there was general consensus during all 3 periods regarding the criteria for active resuscitation in the delivery room. Antenatal steroid therapy to enhance fetal pulmonary maturity was initiated after 1990.13 Neonatal care was rendered according to established guidelines, with no significant changes in the clinical facility or care team during the 3 periods.14,15 Maternal information included age, level of education, race and marital status, duration of gestation, antenatal steroid therapy, and mode of delivery. Gestational age was determined from the date of the last menstrual period and confirmed with obstetric measures, including ultrasonographic findings in the majority of cases. Infant birth data included birth weight, multiple birth status, Apgar scores, and use of assisted ventilation in the delivery room. Neonatal morbidity included respiratory distress, defined as need for oxygen therapy; chronic lung disease, defined as an oxygen dependence at 36 weeks' CA (CA-postmenstrual plus postnatal age)16; patent ductus arteriosus, confirmed with echocardiography; episodes of sepsis, defined as clinical signs of infection with a positive blood culture; necrotizing enterocolitis (NEC)17; intraventricular hemorrhage18; and periventricular leukomalacia. Routine cranial ultrasonographic evaluations were performed according to a similar schedule, with the initial evaluation between day of life 3 to 7 and subsequent evaluations on days of life 10, 30, 60, and before hospital discharge for all study subjects. Severe cerebral ultrasonographic abnormalities were categorized as the presence of either grade III or IV intraventricular hemorrhage, periventricular leukomalacia, or persistent of ventricular dilation at hospital discharge. Specific neonatal therapies included surfactant, duration of oxygen and ventilator therapy, indomethacin and postnatal steroids. Surfactant was used starting in 1990 for infants who required assisted ventilation and ≥30% ambient oxygen to maintain an arterial oxygen pressure of ≥50 mmHg. Postnatal steroid therapy was prescribed at the discretion of the attending neonatologist for infants with chronic lung disease and prolonged ventilator dependence. Indomethacin therapy was used to treat symptomatic patent ductus arteriosus.
Survival rates and neurodevelopmental status were measured to 18 to 20 months' CA. Measures of neurodevelopmental outcomes included neurosensory status and the Mental and Psychomotor Developmental Indices of the Bayley Scales of Infant Development (BSID).19 Children born after 1991 were tested with the revised BSID II, which has been shown to give lower scores than the original BSID used for children born 1982 through 1991. Thus, we devised a correction factor based on the published differences found in a sample of 200 children given both tests in counterbalanced order to compare the scores obtained with the BSID II with those obtained with the original BSID.20 For the Bayley mental development index (MDI), the correction factor ranged from 2 points for a score of 51 to 12 points for a score of 101. For the Bayley psychomotor development index (PDI), the correction factor ranged from 13 points for a score of 51 to 7 points for a score of 101. A neurologic examination of muscle tone was performed by either of 2 trained and certified examiners.21 Cerebral palsy was defined as a persistent but not unchanging disorder of movement and posture appearing in early life and attributable to a nonprogressive disorder of the brain, the result of interference during its development.22 Only patients with moderate-to-severe cerebral palsy were classified as having cerebral palsy for the study. Patients with isolated findings of increased or decreased tone in the extremities were classified as hypertonia or hypotonia on the basis of specific findings. Major neurologic impairments included cerebral palsy, hypotonia, hypertonia, and shunt-dependent hydrocephalus without other neurologic abnormality. Hypotonia and hypertonia were included in the category of major neurologic abnormality, because these conditions are considered by some to represent a variant of cerebral palsy. Shunt-dependent hydrocephalus without neurologic abnormality was also considered an impairment. Neurodevelopmental impairment included any major neurologic impairment, unilateral or bilateral blindness or deafness requiring a hearing aid, and/or an MDI of <70 on the BSID. Consistent criteria for defining neurodevelopmental impairment were used throughout the 21 years of study. Thirty-five neurologically impaired children (4 from period I, 22 from period II, and 9 from period III) were not testable with the BSID because of either behavioral problems or severe impairment. They were included among those with neurologic impairment but who do not have BSID scores reported. The 4 children from period I who had isolated blindness were inadvertently not included in our previous publication on the differences between survival and morbidity rates between the 1980s and 1990s; thus, rates of blindness and overall neurodevelopmental impairment shown for period I in this article were slightly greater than those previously reported.12 Parents provided informed consent for participation in the study, which was approved by the Institutional Review Board of University Hospitals of Cleveland.
ANALYSIS OF DATA
Trends in delivery room care, survival, neonatal morbidity, and outcomes at 20 months' CA were compared between the 3 periods of study. Statistical comparisons were made with unpaired t tests for continuous variables and χ2 analyses for categorical variables. We adjusted for gestational age via logistic regression in the examination of death and survival with and without impairment, because gestational age differed between periods (Table 1). In addition, analysis of variance was performed to compare the mean MDI scores and rates of subnormal MDI over the 3 periods, controlling for gestational age and maternal education.
Comparison Of Perinatal and Birth Data
RESULTS
Characteristics of the Infants, Delivery Care, and Survival Rates
Table 1 presents a comparison of perinatal and birth data between periods I, II, and III. Birth weight was similar, although gestational age decreased significantly over the 3 periods. One infant from period III, who was born with a heartbeat at 17 weeks' gestation but died in the delivery room, was included in the study population because he met the World Health Organization's definition of a live birth.23 Infant race, gender, and number of multiple births did not differ. Antenatal steroid therapy was not used in period I, whereas 41% and 78% of mothers received this therapy in periods II and III respectively (P < .01). Cesarean sections increased significantly over the 3 periods and reached nearly half of all ELBW births during period III (P < .01). During period I, 81% of infants received assisted ventilation compared with 89% during period II (P < .01). There was no additional increase in the rate of assisted ventilation during period III. Surfactant was administered to only 1% of infants during period I, whereas 62% and 82% of infants received surfactant during periods II and III, respectively (P < .01).
A total of 242 (49%) of 496 infants survived to 20 months' CA during period I, compared with 508 (68%) of 749 during period II (P < .001). There was no significant change in survival between periods II and III at which time 165 (71%) of 233 infants survived. Compared with periods II and III, survivors from period I were of significantly higher birth weight (826 g vs 800 and 801 g; P < .01) and gestational age (26.7 weeks vs 26.2 and 26.0 weeks; P < .001).
Cause of Death
The most common cause of death was respiratory distress syndrome, accounting for approximately one third of all infants who died during the 3 periods. (Table 2) Another one third of the infants were not offered assisted ventilation because they were not considered viable; therefore, the cause of death was defined as “immaturity.” NEC as the primary cause of death occurred significantly more often during period III (P < .01). The increased mortality from NEC is unexplained in light of the overall reduction in sepsis during period III. During period I, 67% of deaths occurred before age 48 hours versus 58% and 57% during periods II and III, respectively (P = .06).
Primary Cause and Age of Death
Neonatal Morbidity of the Survivors
Chronic lung disease increased significantly from period I to II (P < .01) but did not change between period II and III (P = .33) (Table 3). Postnatal corticosteroid therapy, which was rarely used before 1990, increased dramatically in period II then decreased by nearly half in period III. The rate of sepsis also increased from period I to II (P < .001) then decreased significantly in period III (P < .001). The rate of grades III and IV intraventricular hemorrhage decreased between period II and III (P = .004). Periventricular leukomalacia increased between periods I and II (P = .018), likely because of improved diagnostic techniques, but did not change between period II and III. The overall rate of severe cranial ultrasound abnormalities decreased during period III (P < .001).
Neonatal Morbidity Among Surviving Infants of 500- to 999-g Birth Weight
Neurodevelopmental Outcomes at 20 Months' CA
Complete information on neurodevelopmental status at 20 months' CA was available for 218 (90%) survivors born during period I and 467 and 152 survivors (both 92%) born during periods II and III, respectively. Children who were followed had similar birth weights and gestational ages compared with those lost to follow-up. The educational level of the mothers was higher during period II, with 83% (417 of 500 subjects) having completed high school versus 77% (181 of 236 subjects) and 76% (123 of 162 subjects) during periods I and III, respectively (P = .03). Maternal race (64%, 61%, and 57% black) and marital status (56%, 54%, and 59% unmarried) did not differ between periods.
The overall rate of neurologic abnormalities was highest during period II (Table 4). Cerebral palsy, which increased from 8% in period I to 13% in period II, decreased to 5% in period III (P = .008). Blindness decreased between periods I and II, likely because of improvements in laser therapy for severe retinopathy of prematurity, but did not change between periods II and III. Deafness, which doubled between period I and II, decreased during period III.
Neurodevelopmental Outcomes at 20 Months' CA
There were no significant differences in the mean MDI or in the rates of subnormal MDI scores (<70) during the 3 periods (Table 4). However, the mean PDI was significantly lower in period II, most likely attributable to the higher rate of cerebral palsy during this period. The overall rates of neurodevelopmental impairment, including neurosensory abnormalities and/or subnormal MDI scores, increased from period I to II (28% vs 35%; P = .017), then decreased significantly to 23% in period III (P = .01). For children without neurosensory abnormality, the mean MDI scores were 90.4, 88.5, and 88.5, respectively, and rates of subnormal MDI (<70) were 12%, 14%, and 15% for the 3 periods, respectively. These did not differ between periods. Analysis of variance, controlling for gestational age and maternal education revealed no significant differences in mean MDI or rates of subnormal MDI over the 3 periods for all ELBW children and for the neurosensory normal subgroup.
Figure 1 illustrates the outcomes of all live-born infants. From period I to II, mortality decreased, and survival both with and without neurodevelopmental impairment increased significantly. The overall rates of death and/or impairment decreased from 67% to 57% (P < .01). Mortality did not change between periods II and III, but intact survival increased, whereas neurodevelopmental impairment decreased. The overall rate of death and/or impairment further decreased to 47% (P < .01).
Comparison of death and survival with and without neurodevelopmental impairment at 20 months' CA for 500- to 999- g birth weight infants born during 3 periods: 1982–1989; 1990–1999, and 2000–2002. a Neurosensory abnormality and/or MDI <70.
In a comparison of periods I and II, after adjustment for gestational age, period II had a lower death rate, odds ratio (odds ratio [OR]: 0.4; 95% confidence interval [CI]: 0.3–0.6), a higher rate of survival with impairment (OR: 2.0; 95% CI: 1.5–2.8), and a higher rate of survival without impairment (OR: 1.5; 95% CI: 1.2–1.9). A similar comparison between periods II and III yields an OR for death of 0.9 (95% CI: 0.6–1.2); for survival with impairment, the OR was 0.6 (95% CI: 0.4–0.96); and for survival without impairment, the OR was 1.5 (95% CI: 1.1–2.0).
To determine the relative effect of birth period and various perinatal and neonatal factors contributing to increased survival without neurodevelopmental impairment, logistic regression analysis was performed for patients from study periods II and III who survived to age 7 days, controlling for birth weight, gender, study period, and significant factors in the univariate comparison including antenatal steroid use, mode of delivery, surfactant therapy, severe cranial ultrasound abnormality, sepsis, and postnatal steroid exposure. Patients who survived < 7 days were omitted from the analysis because they frequently lacked cranial ultrasound data and were often not eligible for postnatal steroid therapy. After adjusting for the significant factors, neurodevelopmental impairment-free survival did not differ between periods II and III (OR: 0.93; 95% CI: 0.59–1.46; P = .76). Detailed examination of these results suggested that the lower rates of impairment during period III seen in the unadjusted analyses might be explained by significant decreases in postnatal steroid therapy, severe cranial ultrasound abnormality, and sepsis. These 3 factors were all significant predictors of neurodevelopmental impairment-free survival in the multivariate logistic model. ORs and 95% CIs for impairment-free survival were 0.59 (0.41–0.85) for postnatal steroid therapy; 0.16 (0.11–0.25) for severe cranial ultrasound abnormalities and 0.64 (0.45–0.89) for sepsis, with P values of <.01, <.001, and 0.01, respectively. In contrast, a protective effect for antenatal steroids on intact survival bordered on significant (OR: 1.37; 95% CI: 0.96–2.0; P = .09). The same multivariate logistic model to predict neurodevelopmentally intact survival using significant antenatal and birth factors such as study period, birth weight, gender, cesarean section delivery, antenatal steroid treatment, and surfactant therapy was performed for the entire population of infants born during periods II and III. In this analysis, postnatal factors, such as severe cranial ultrasound abnormalities, sepsis, and postnatal steroid treatment were not included because of the lack of opportunity for exposure to these factors for infants who died before 7 days age. In this analysis, the protective effect of antenatal steroids on intact survival reached significance (OR: 2.01; 95% CI: 1.48–2.74; P < .001). Cesarean section delivery and female gender were also significantly protective (OR: 1.8; 95% CI: 1.35–2.46; P < 0001 and OR: 2.2; 95% CI: 1.62–2.94; P < .001, respectively). Surfactant did not have a protective effect on intact survival in either multivariate logistic analysis. Of interest, multivariate analysis using both prenatal and postnatal factors performed for infants surviving to follow-up suggests that although antenatal steroids, female gender, and cesarean section delivery enhanced survival, severe cranial ultrasound abnormalities, sepsis, and postnatal steroid therapy predicted neurodevelopmental outcome.
DISCUSSION
Our results indicate that although survival rates for ELBW infants born at our perinatal center increased during the 1990s compared with the 1980s, no additional increase in survival occurred during 2000–2002. During the 1990s, the perinatal approach to ELBW infants became more aggressive with increased use of antenatal steroids, cesarean section delivery, surfactant therapy, assisted ventilation, and postnatal steroids. The associated neonatal morbidities increased, resulting in higher rates of neurodevelopmental impairment. Between 2000 and 2002, antenatal steroid use continued to increase, but postnatal steroid use declined dramatically after reports of increased rates of cerebral palsy.24–28 Despite this decrease in postnatal steroid use, there was no change in the rate of chronic lung disease during period III. At the same time, rates of sepsis and severe intraventricular hemorrhage also decreased. Neurodevelopmental outcomes improved for ELBW infants born during 2000–2002, with decreased rates of cerebral palsy, deafness, and overall neurodevelopmental impairment. During the 1990s, there was increased survival both with and without impairment. Since the year 2000, there have been improved neurodevelopmental outcomes, with increased survival without impairment and a decrease in survival with neurodevelopmental impairment. A variety of perinatal and neonatal factors were responsible for the improved outcomes, including decreases in postnatal steroid therapy, severe cranial ultrasound abnormalities, and sepsis.
Increasing rates of neurodevelopmental disability among ELBW infants born in the early 1990s have been previously reported.12,29,30 However, others suggest stable or slightly improved neurodevelopmental outcomes associated with increased antenatal steroid use.31,32 Our study, which is the first, to our knowledge, to report improved outcomes for 2000–2002 in conjunction with decreased use of postnatal steroids, suggests that the adverse neurodevelopmental outcomes during the 1990s likely relate to postnatal corticosteroid use and sepsis. Dexamethasone, the glucocorticoid most commonly used in the clinical setting, has been shown to impair cortical growth among preterm infants.33–35 Research has demonstrated that the preferential binding of dexamethasone to the glucocorticoid receptors of the dentate gyrus induces apoptosis and neuronal cell death.36
Several authors have noted an association between sepsis and cerebral palsy.37–40 A recent report of ELBW infants identifies sepsis as an independent risk factor for neurodevelopmental impairment, including cerebral palsy and subnormal cognition.41 Inflammatory cytokines, released during intrauterine infection, are considered to be catalysts for the development of periventricular leukomalacia and subsequent cerebral palsy.42,43 Recent neonatal quality improvement initiatives have introduced practice guidelines to minimize the risk of nosocomial sepsis.44 Our neonatal unit now emphasizes aseptic precautions for all central line sites, use of in-line blood sampling equipment, and meticulous hand-washing. Such policies together with restriction of postnatal steroid therapy, which is associated with increased susceptibility to infections, may have contributed to our improved neurodevelopmental outcomes.45,46
The strengths of our study include the relatively large inborn population and the high follow-up rates. The major limitation concerns the relatively short-term nature of the follow-up. Children diagnosed with cerebral palsy may “outgrow” the diagnosis in later years,47 and cognitive scores may improve later in childhood.48,49 The recent 3-year observational study period may not provide sufficient time to fully reveal the impact of therapeutic interventions on outcome. The presentation of outcomes by birth weight rather than gestational age will include small-for-gestational-age infants who are likely more mature. This study represents a predominantly urban population. Thus, its findings may not be generalizable to other neonatal centers with larger numbers of suburban infants.
Survival did not increase between periods II and III, as has also been shown by others.1,50 Because additional reduction in the mortality rates of ELBW infants seems unlikely, future efforts must concentrate on reducing neonatal morbidity. Although the decrease in cerebral palsy since 2000 is encouraging, the lack of improvement in MDI is unexplained and necessitates additional research for interventions that may enhance cognitive function. Sociodemographic and environmental factors may contribute to differences in cognitive outcomes. Although maternal education was significantly higher during period II, analysis of variance controlling for maternal education did not change our outcomes. Whereas some studies on the effects of early educational intervention programs have demonstrated improved outcomes for children born to mothers of low sociodemographic status,51,48 other reports have failed to demonstrate a sustained benefit.52,53 The use of inhaled nitric oxide as rescue therapy to reduce death or chronic lung disease among preterm infants with respiratory failure has also been shown to decrease the rates of intraventricular bleeds and improve cognitive outcomes.54,55 However, these findings have been tempered by studies that suggest either no improvement or worsened outcomes among ELBW infants.56,57 Meanwhile, the iatrogenic detrimental effects of postnatal steroids caution against the premature use of new therapies for the most vulnerable infants before a thorough evaluation by randomized, controlled clinical trials.58
CONCLUSIONS
Advances in perinatal care before 2000 resulted in improved survival but with either an increase or lack of reduction in neurodevelopmental impairment.2,12 Greater numbers of disabled survivors result not only in enormous family impact but also the need for additional resources for support in school, the home, and workplace. The decreased morbidity and impairment that we have described since 2000 is a step in the right direction. However, as resuscitation and neonatal care continue to save the lives of the smallest and least mature infants, additional advances will be required to increase survival free of neonatal morbidity and neurodevelopmental impairment. Longer term follow-up of this population will be essential to evaluate the effects on quality of life for infants and their families and to help determine what constitutes appropriate care without doing harm. At the present time, prematurity cannot be prevented.59 Emphasis on decreasing the number of multiple births resulting from infertility therapies accompanied by minimizing iatrogenesis and morbidity associated with sepsis and postnatal steroid therapy may hold promise for improved long-term outcomes.
Acknowledgments
This work was supported by grant M01RR00080, General Clinical Research Center, and partially supported by grant HD21364 from the National Institute of Child Health and Human Development Neonatal Research Network.
We thank Angelia Williams and Nancy Newman, RN, for help in conducting this research.
Footnotes
- Accepted September 25, 2006.
- Address correspondence to Deanne Wilson-Costello, MD, Division of Neonatology, Rainbow Babies & Children's Hospital, University Hospitals of Cleveland, 11100 Euclid Ave, Cleveland, OH 44106-6010. E-mail: drfjcmd{at}aol.com
The authors have indicated they have no financial relationships relevant to this article to disclose.
REFERENCES
- Copyright © 2007 by the American Academy of Pediatrics