Neurologic Outcomes at School Age in Very Preterm Infants Born With Severe or Mild Growth Restriction
OBJECTIVE: To determine whether mild and severe growth restriction at birth among preterm infants is associated with neonatal mortality and cerebral palsy and cognitive performance at 5 years of age and school performance at 8 years of age.
METHODS: All 2846 live births between 24 and 32 weeks' gestation from 9 regions in France in 1997 were included in a prospective observational study (the EPIPAGE [Étude Epidémiologique sur les Petits Ages Gestationnels] study) and followed until 8 years of age. Infants were classified as “small-for-gestational-age” (SGA) if their birth weight for gestational age was at the <10th centile, “mildly-small-for-gestational-age” (M-SGA) if birth weight was at the ≥10th centile and <20th centile, and “appropriate-for-gestational-age” (AGA) if birth weight was at the ≥20th centile.
RESULTS: Among the children born between 24 and 28 weeks' gestation, the mortality rate increased from 30% in the AGA group to 42% in the M-SGA group and to 62% in the SGA group (P < .01). Birth weight was not significantly associated with any cognitive, behavioral, or motor outcomes at the age of 5 or any school performance outcomes at 8 years. For the children born between 29 and 32 weeks' gestation, SGA children had a higher risk for mortality (adjusted odds ratio [aOR]: 2.79 [95% confidence interval (CI): 1.50–5.20]), minor cognitive difficulties (aOR: 1.73 [95% CI: 1.12–2.69]), inattention-hyperactivity symptoms (aOR: 1.78 [95% CI: 1.10–2.89]), and school difficulties (aOR: 1.74 [1.07–2.82]) compared with AGA children. Being born M-SGA was associated with an increased risk for minor cognitive difficulties (aOR: 1.87 [95% CI: 1.24–2.82]) and behavioral difficulties (aOR: 1.66 [95% CI: 1.04–2.62]).
CONCLUSIONS: In preterm children, growth restriction was associated with mortality, cognitive and behavioral outcomes, as well as school difficulties.
- small for gestational age
- neurological outcome
- cognitive performance
- cerebral palsy
- school performance
WHAT'S KNOWN ON THIS SUBJECT:
The association between SGA and neurodevelopmental outcome in the preterm infant is not well defined, with variable and evolving definitions of SGA.
WHAT THIS STUDY ADDS:
We investigated the consequence of classic SGA (<10th percentile) and also a category called mild-SGA (10th–19th centile) (M-SGA) in the premature population. Both SGA and M-SGA were associated with mortality and with cognitive, behavioral, and school difficulties.
Over the last few decades, improvements in the perinatal management of premature newborn infants have made it important to consider the long-term outcome of these infants. Recent studies have found that preterm birth remains associated with high rates of severe long-term neurodisabilities (cerebral palsy, mental retardation, and sensory impairments), and even without severe impairments many children have serious developmental lags.1 Encephalopathy of prematurity is a new concept that is used to explain the neurologic sequelae of the preterm.2 It is defined by a destructive disease with white matter injury and a neuronal/axonal disease associated with maturational and trophic disturbance, which lead to serious neurodevelopmental disabilities including cerebral palsy and cognitive and behavioral deficits. Being preterm and small for gestational age (SGA) may represent a double risk, because both of them are associated with adverse outcomes.3,4
Studies have shown conflicting results regarding the association between SGA and neurodevelopmental outcomes in the premature population. SGA at term has been associated with poorer cognitive abilities in some studies.5 However, other studies have not found such an association, especially for preterm infants.6,7 Although behavioral problems in relation to preterm birth and/or SGA8 have been studied in young children9 or teenagers,5,10 subtle neurologic impairments, such as mild cognitive dysfunction, behavioral problems, and school achievement, have been poorly studied in children at school age, specifically in children who were born preterm and SGA.
SGA is commonly defined as a birth weight for GA at the <10th centile. However, growth restriction is a dynamic process, and being at the <10th centile at 25 weeks' gestation is different from being at the <10th centile at 31 weeks or at term. Furthermore, because preterm infants are not born after “normal” pregnancies and “normal” fetuses are not born prematurely, growth references that use preterm live births underestimate growth restriction.11,12 Growth restriction in preterm infants requires a less strict cutoff than the 10th centile and needs to be investigated according to GA. In this way, the consequences of growth restriction should also be studied according to its severity and according to a given GA.13
In this study, we investigated the influence of SGA on motor, cognitive, and behavioral outcomes in extremely preterm (24–28 weeks' gestation) and very preterm (29–32 weeks' gestation) children at the ages of 5 and 8 years. We determined whether the “classic” SGA (<10th centile) and a new category called mildly SGA (M-SGA) (between the 10th and 19th centiles) were associated with different neurodevelopmental outcomes.
POPULATION AND METHODS
The children of our study were part of the Étude Epidémiologique sur les Petits Ages Gestationnels (EPIPAGE) cohort, which included all births before 33 completed weeks of amenorrhea in all maternity units, regardless of the level of the hospital, of 9 regions of France between January 1 and December 31, 1997. These children were followed from birth to the ages of 5 and 8 years. All survivors were born between 24 and 32 completed weeks of amenorrhea. A term reference group (666 children) was included at birth in the same regions (1 in every 4 births at 39 or 40 weeks of amenorrhea during 1 week). In the preterm group, 9 children had no data on birth weight or gender and were excluded (Fig 1). The study population comprised 2846 preterm live births. Of these infants, 2458 (86%) were discharged from the hospital. In 2 regions, half of the infants born at 32 weeks were randomly excluded from the follow-up to reduce the workload. This process resulted in follow-up of 2382 infants. There were 25 deaths before 5 years, which left 2357 children for follow-up. At 5 years of age, each child was invited to be examined by a pediatrician who performed a detailed neurologic examination. A psychologist performed a test of cognitive abilities, and parents were requested to complete a questionnaire on the behavior of their child. Of the 2357 preterm children who survived and who were eligible for follow-up, 1812 (77%) children had a medical examination, 1535 (65%) had a complete psychological test, and 1677 (71%) had a behavioral assessment. At 8 years, a parental questionnaire was sent to assess school achievement. This questionnaire was available from 1439 (61%) of the preterm children who survived and were eligible for follow-up at this age.
Informed verbal consent from the parents was obtained by the medical team in charge of the study. This study was approved by the French data-protection agency.
Pregnancy and Neonatal Data
Gestational age (GA) referred to completed weeks of amenorrhea, which was the best obstetric estimate and combined last menstrual period and early prenatal ultrasound and clinical assessments. The following social and demographic characteristics were recorded at birth: mother's age (<25, 25–34, or ≥34 years), social class of the family (parents' highest level of occupation or the mother's highest level of occupation if she lived alone), mother's nationality (French or other), and parity (0, 1–2, ≥3). Perinatal characteristics were extracted from medical records and included gender, multiple births, and antenatal steroids.
Centiles of birth weight for GA were estimated by weeks of amenorrhea and gender by using our cohort of very preterm live births. Infants were classified as SGA if their birth weight for GA was at the <10th centile of the population. Accordingly, the other infants were classified as M-SGA if their birth weight for GA was at the ≥10th centile and <20th centile and appropriate for GA (AGA) if their birth weight for GA was at the ≥20th centile.
Hospital mortality was defined as death in a maternity ward or neonatal unit.
At 5 years of age, each child was subjected to a detailed medical and neurologic examination in which tone, reflexes, postures, and movements were assessed. Trained pediatricians reviewed data for children with abnormal results on neurologic examination to validate the diagnosis of cerebral palsy and to assess its severity. We used the European Cerebral Palsy Network definition of cerebral palsy.14 Children were classified as having cerebral palsy if they had abnormal posture or movement, increased tone or hyperreflexia (spastic cerebral palsy), involuntary movements (dyskinetic cerebral palsy), or loss of coordination (ataxic cerebral palsy).
The French version of the Kaufman Assessment Battery for Children,15 administered by a trained psychologist, was used to assess cognitive function. We used the Mental Processing Composite (MPC) scale, which has been standardized to have a mean of 100 and an SD of 15 in a French population born in the 1990s.16 Cognitive deficiency was defined as an MPC score <85 (–1 SD).
Behavioral problems were assessed with the French version of the Strength and Difficulties Questionnaire,17 which was completed by the parents. It includes 25 items from 5 rating scales (hyperactivity-inattention, conduct, emotional and peer problems, and prosocial behavior). Scores for the first 4 scales are summed up in a “total difficulties” score, which indicates poorer mental health. For each item, parents can choose between 3 answers: not true; moderately true; and certainly true. Cutoffs were defined so that 10% of the term control group in the EPIPAGE cohort was considered to have a behavioral problem.18
At 8 years of age, a questionnaire was sent by mail to the parents to obtain data on school performance. School difficulties were defined by special schooling (institution or special school, special class in mainstream school, mainstream class) or low grades.
We studied bivariate relationships between birth weight for GA and each outcome (ie, hospital mortality, cerebral palsy, cognitive deficiency, behavioral problems, and school achievement) by using χ2 or Fischer's exact test for categorical variables. Logistic regression models were used to study these associations after adjustment for potential confounders and GA in weeks. Covariates were included in the model if they were known risk factors and found to be associated with the studied outcome at the 20% significance level in univariate analysis (see Appendix). Analyses were conducted on 2 strata of GAs (24–28 and 29–32 weeks) to determine if associations between each outcome and birth weight for GA differed according to the GA. In each strata, we adjusted for GA in weeks. Multivariate associations between birth weight for GA and each outcome were studied by using Wald tests, and results are expressed as adjusted odds ratios (aORs) and their 95% confidence intervals (95% CIs). All statistical tests were 2-sided, and P < .05 was regarded as significant.
Statistical analysis was performed by using Stata 10.0 (Stata Corp, College Station, TX).
Among the 2846 infants born alive, 274 (9.6%) were M-SGA and 262 (9.2%) were SGA; 828 children were born between 24 and 28 weeks' gestation, and 2018 were born between 29 and 32 weeks' gestation (Table 1). Among children in the 24- to 28-week group there was no significant association between weight for GA and social and maternal variables (Table 2). In the 29- to 32-week group, the proportions of SGA and M-SGA were increased among nulliparous women older than 35 years and among women who received antenatal corticosteroids (Table 2).
In the 24- to 28-week group (Table 3), the mortality rate increased significantly (P < .01) as the GA for birth weight decreased and ranged from 30% in the AGA group, 42% in the M-SGA group, and 62% in the SGA group. After adjustment, both the SGA (aOR: 5.07 [95% CI: 2.72–9.46]) and M-SGA (aOR: 1.78 [95% CI: 1.00–3.18]) groups had a significantly increased mortality rate compared with the AGA group. No significant associations were found between birth weight for GA and cerebral palsy (P = .75), cognitive deficiency (P = .85), behavioral problems (P > .5), or school difficulties (P = .43).
In the 29- to 32-week group (Table 4), infants born SGA, but not M-SGA, had a significantly increased risk of death compared with infants born AGA (aOR: 2.79 [95% CI: 1.50–5.20]). Although children born SGA or M-SGA had a lower rate of cerebral palsy at 5 years of age than children born AGA, these associations were not significant. In contrast, >40% of infants born M-SGA or SGA had an MPC score of <85 vs 29% of the infants born AGA (P < .01). After adjustment, the associations between an MPC score of <85 and the SGA (aOR: 1.73 [95% CI: 1.12–2.69]) and M-SGA (aOR: 1.87 [95% CI: 1.24–2.82]) group scores remained significant. Birth weight for GA was significantly associated with inattention-hyperactivity symptoms and total behavioral problems among these children. After adjustment, children born SGA remained at increased risk of inattention-hyperactivity symptoms (aOR: 1.78 [95% CI: 1.10–2.89]), whereas children born M-SGA remained at increased risk of total behavioral difficulties (aOR: 1.66 [95% CI: 1.04–2.62]). No significant associations were found between birth weight for GA and behavioral, emotional, or peer problems or prosocial behavior (data not shown). At 8 years of age, the frequency of school difficulties was significantly different between children born SGA (28%), M-SGA (23%), and AGA (18%) (P = .04). After adjustment, school difficulties were associated with being born SGA (aOR: 1.74 [95% CI: 1.07:2.82]).
With this large cohort study we found that being born SGA or M-SGA was associated with greater risk of mortality in both GA groups than being born AGA. Growth restriction was associated with adverse neurodevelopmental outcomes only in the 29- to 32-week GA group. Within this group, SGA, as commonly defined in clinical practice (<10th centile), was associated with an increase of cognitive deficiency, behavioral problems, and school difficulties. Although infants born M-SGA (10th–19th centile) are not usually considered growth restricted, children in this group were at risk of impairments such as cognitive deficiency and behavioral problems.
The strengths of this study are the prospective design and the long-term follow-up (8 years). Inclusions were based on GA, which was estimated by the best available techniques (ultrasound scan results and date of last menstrual period). Because an early scan is a standard practice for almost all pregnant women in France,19 we can assume that the estimation of GA in this study was reliable and of good quality. Furthermore, the population of the EPIPAGE study was recruited on a geographical basis, which reduced the bias that is inherent to studies that select their patients from chosen perinatal centers. We used validated questionnaires to define cerebral palsy,14 cognitive impairment,16 and behavioral difficulties.17 Socioeconomic characteristics known to be associated with both preterm birth and cognitive and behavioral outcome were collected in our study,19 which made it possible to control for these factors in multivariate analyses.
Loss to follow-up is an important issue for longitudinal studies. In our study, information was available for 65% and 61% of the eligible population at 5 and 8 years, respectively. These rates of follow-up have to be considered in light of the large number of children included and their geographic dispersion. However, several studies have found that children lost to follow-up are more likely to have a disability.20 In our study, there was no difference between participants and nonparticipants at 5 and 8 years of age according to GA at birth, birth weight for GA, and cerebral lesions at neonatal cranial ultrasound examination. Mean GA was 29.9 weeks for participants versus 30.1 weeks for nonparticipants (P < .01). However, the nonparticipating families were more frequently from lower social classes and younger, multiparous, and foreign-born mothers than those who participated (data not shown), which could have led to an underestimation of cognitive and behavioral difficulties, which are known to be associated with social environment. Whether this selection induced a bias in the associations observed between birth weight for GA and long-term outcome is less clear.
We used a neonatal internal reference to approach in utero growth restriction, because few data were available in France to determine the appropriate weight at low GA, especially before 28 weeks' gestation. The growth curves that are routinely used in France are those of Mamelle et al.21 It is unfortunate that few data are available for preterm children born before 28 weeks except for foreign curves.22 In our study, for children born at 29 to 32 weeks' GA, the 10th and 20th centiles were similar to those of Mamelle et al.23 It would have been of interest to define the birth weight for GA by using the target birth weight.24 However, insufficient data were available on maternal height and weight to allow the use of this strategy, and the available data concern mainly term-born children.24 Therefore, some infants were misclassified, but these errors should have underestimated our results.
Preterm infants are smaller than their in utero peers,25 and the true proportion of growth-restricted newborn infants is underestimated when the definition is based on the 10th centile for weight on a neonatal growth curve.11 We used an intermediate class of centiles (ie, 10th–19th) to define infants born M-SGA. This group had increased risks of mortality and cognitive and behavioral impairment. Furthermore, the consequences of abnormal fetal growth seem to be a function of etiology, severity, timing, and pregnancy duration.26 Extremely preterm and very preterm infants are affected differentially by growth restriction, which is not well reflected in the current SGA classification scheme when using the 10th centile for each gestational strata.
The 24- to 28-week GA group was characterized by a high mortality rate (from 30% to 62%). Although this study was conducted in 1997, antenatal steroids, early surfactant administration, and new ventilation techniques were already widely used. Because recruitment was geographically based, our result reflected the general practice during the study and not that of selected neonatology centers. In this population, 50% of deaths in the delivery room or the NICU occurred after a decision to limit intensive care.27 This practice had 2 consequences: first, the survivors in the 24- to 28-week GA group were highly selected; second, the small number of survivors limited the statistical power to investigate the association between the birth weight for GA and developmental outcomes. Another hypothesis could be that severe immaturity and its consequences have stronger effects than growth restriction at these early GAs.
We did not find any association between birth weight for GA and cerebral palsy. The authors of a large European study reported a significantly increased risk of cerebral palsy in infants born SGA after 32 weeks' gestation but not in those born before this GA,28 which is probably because of the competition between cerebral palsy and the increased risk of mortality in this group. In our study, the absence of association between SGA and cerebral palsy should also be interpreted with caution, because children born AGA were born after various pregnancy complications (ie, preterm premature rupture of membranes, idiopathic preterm labor, or maternal hemorrhage). Among them, those who were exposed to prenatal infection (15% of children born after idiopathic preterm labor and 35% of those born after preterm premature rupture of membranes29,30) are known to have a high risk of white matter damage and subsequent cerebral palsy.31 Thus, this control group had a heterogeneous risk of cerebral injury and cerebral palsy, which made it difficult to interpret results of the comparison for SGA children.
Another finding of our study was that increased risk of cognitive deficiency was significantly associated with mild and severe growth restriction in the 29- to 32-week GA group. This result was inconsistent with previous studies that did not find such association at 24 months of age.32 This discrepancy may be because mild-to-moderate disabilities are identified later than more severe problems. Later-emerging sequelae only become apparent when the demands for higher-level skills cannot be met because of the underlying subtle cognitive difficulties. Although associations with long-term outcomes have been reported from several studies,5 they have been restricted to the impact of severe SGA (ie, 3rd or 10th centile). One hypothesis that was recently proposed to explain the associations between cognitive deficiency and growth restriction is hypoxia caused by placental insufficiency that affects the gray matter33 despite brain-sparing.34
Although it has been reported that there is a high incidence of attention-deficit disorder and behavioral difficulties among low birth weight and/or preterm children,35,36 SGA status in preterm infants was not known to have independent effects on attention.5 Furthermore, according to Yanney and Marlow,37 the effects of prematurity seem to overcross the effects of fetal growth restriction at GAs of <32 weeks. Our findings that SGA increased the risk of inattention-hyperactivity symptoms and that M-SGA increased the risk of total behavioral disorders at 5 years of age, therefore, are of particular interest. In the Helsinki study conducted in young adults born prematurely,38 SGA was associated with behavioral difficulties such as executive dysfunction and emotional instability but not with inattention. Because the prevalence of attention-deficit/hyperactivity symptoms decreases with age, the absence of association between SGA and inattention in young adults could be expected,39 which suggests a delayed development in these domains among preterm infants born SGA rather than a permanent cerebral dysfunction. However, behavioral characteristics are likely to be influenced also by a broad range of genetic/epigenetic and nurture factors. Thus, these results should be interpreted with caution.
Learning difficulties, lower school grades, and the need for special education are more frequent in preterm children than in the general population. Our results indicate that in the population of preterm infants born between 29 and 32 weeks' gestation, children born SGA had significantly increased educational disadvantages compared with their AGA peers. This same trend of educational difficulties at school has been reported by other authors.40,41
To our knowledge, our study was the first to investigate the short-term and long-term consequences of severe and mild growth restriction among very preterm children. We observed that specific disabilities were associated with mild growth restriction, which is rarely considered as a risk factor for neurodevelopmental outcome. These results suggest that children with growth restriction, including mild growth restriction, could benefit from neurologic assessment until school age and perhaps from early developmental interventions. Additional research is needed to study how the postnatal growth of these children may intervene with these features.
- Accepted January 18, 2011.
- Address correspondence to Isabelle Guellec, MD, Réanimation Neonatale et Pédiatrique, Trousseau Hospital, Boulevard A. Netter, 75012 Paris, France. E-mail: or
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
COMPANION PAPER: Companions to this paper can be found on pages e874 and e1048 and online at www.pediatrics.org/cgi/doi/10.1542/peds.2010-1821 and www.pediatrics.org/cgi/doi/10.1542/peds.2011-0262, respectively.
- GA =
- gestational age •
- SGA =
- small for gestational age •
- M-SGA =
- mildly small for gestational age •
- EPIPAGE =
- Étude Epidémiologique sur les Petits Ages Gestationnels •
- AGA =
- appropriate for gestational age •
- MPC =
- Mental Processes Composite •
- aOR =
- adjusted odds ratio •
- CI =
- confidence interval
- McCarton CM,
- Wallace IF,
- Divon M,
- Vaughan HG Jr.
- O'Keeffe MJ,
- O'Callaghan M,
- Williams GM,
- Najman JM,
- Bor W
- Karagianni P,
- Kyriakidou M,
- Mitsiakos G,
- et al
- Aarnoudse-Moens CS,
- Weisglas-Kuperus N,
- van Goudoever JB,
- Oosterlaan J
- Feldman R,
- Eidelman AI
- Hille ET,
- Weisglas-Kuperus N,
- van Goudoever JB,
- et al.
- Hutcheon JA,
- Platt RW
- Hutton JL,
- Pharoah PO,
- Cooke RW,
- Stevenson RC
- Kaufman A,
- Kaufman N
- Kaufman A,
- Kaufman N
- Delobel-Ayoub M,
- Arnaud C,
- White-Koning M,
- et al.
- Keen DV,
- Pearse RG
- Larroque B,
- Bréart G,
- Kaminski M,
- et al.
- Gortner L,
- van Husen M,
- Thyen U,
- Gembruch U,
- Friedrich HJ,
- Landmann E
- Ross G,
- Lipper EG,
- Auld PA
- Saigal S,
- Pinelli J,
- Hoult L,
- Kim MM,
- Boyle M
- Copyright © 2011 by the American Academy of Pediatrics