Article Text

Download PDFPDF

Neurocognitive outcome following fetal growth restriction
  1. D-M Walker,
  2. N Marlow
  1. 1
    Academic Division of Child Health, School of Human Development, University of Nottingham, Nottingham, UK
  1. Professor N Marlow, Department of Child Health, Level E East Block, Queen’s Medical Centre, Nottingham NG7 2UH, UK; neil.marlow{at}


This review evaluates neurological and cognitive outcomes as they relate to fetal growth restriction. The aim is to clarify the relationship between poor fetal growth and abnormal brain development in order to understand the research directions required to untangle this complex issue.

Statistics from

Inthe paediatric literature, many studies confuse the outcome for children who are small for their gestational age (SGA) with the outcome following fetal growth restriction (FGR). Some members of the normal population will be small at birth: by definition, as part of the normal distribution, 10% will have a birth weight below the 10th centile. Often secondary to maternal genetic influences, ethnic group or parity, children who are “small normals” can be identified by the use of customised growth charts,1 2 and there is little evidence to suggest that the long-term outcome for these children is likely to be impaired. Current evidence suggests that standard growth charts based on birth weight, usually derived from live births, may underestimate the degree of deviation from fetal growth, as children born preterm tend to be smaller at birth compared with fetal weight estimates.

Investigators attempting to access children with FGR may try to overcome this by taking an extreme birth weight (eg, below the 3rd or 0.4th centile for gestation, where the “small normal” baby will be less common), but studies in this area would be much improved if they reported outcomes for children where there was good antenatal evidence of failure to thrive, irrespective of where their birth weight fell, or at least used customised growth charts to identify children who had failed to reach their fetal growth potential. Indeed there is a pressing need to understand the influence of poor fetal growth on longer-term outcomes and the extent to which antenatal assessment may predict later outcome or influence perinatal illness.

True FGR is the second leading cause of perinatal morbidity and mortality, exceeded only by prematurity,3 4 and has been reported to occur in about 5% of the general obstetric population.5 The degree of compromise in fetal health during FGR may be determined from changes in regional fetal growth, behaviour and cardiotocography and from studies of umbilical and regional fetal flow velocities using Doppler ultrasound; importantly, these changes may be related to the severity of the perinatal illness.6 Critically, this means that the policy of delivery may have huge implications for longer-term outcomes. If babies are delivered only when there is severe fetal compromise with poor cardiotocographic variability, then their neonatal course may be complicated by a range of issues relating to fetal hypoxia superimposed on prematurity with respiratory failure, hypotension and brain injury, all of which may affect long-term outcomes. In contrast, delivery before fetal health deteriorates may run the risk of complications of very preterm birth but, paradoxically, lead to a less complex neonatal course and better outcomes.

Thus trying to interpret outcome without detailed information on the antenatal and perinatal history is at best likely to be misleading. There are lamentably few attempts to study the effect of timing of delivery on fetal or neonatal outcome in a structured fashion. Indeed only the Growth Restriction Intervention Trial (GRIT) has attempted to do this in the context of a randomised trial, with uncertain results.7 Despite criticism of the generalisability of this study,8 obstetricians continue to deliver babies after FGR with little idea of how their management determines the long-term consequences. In this context, we have reviewed the literature pertaining to long-term outcomes following FGR.


FGR may be characterised by poor intrauterine growth velocity after serial ultrasound estimates during pregnancy in which decreased growth is detected, or birth weight using individualised or customised growth standards and postnatal anthropometric measures. FGR is often further defined as “symmetrical”, where the baby shows early-onset growth restriction on all three variables (ie, weight, length and head circumference), and “asymmetrical”, where the baby appears long and thin and usually shows late-onset growth restriction during gestation, indicating restriction during the period of fat deposition.9 Such a distinction may not be wholly useful, as there is evidence that fetal brain growth slows before growth of the cranial vault,10 but there may be useful long-term clinical correlates.11 12 This failure of brain growth before skull growth may also imply that fetal assessment using biparietal diameters many not be reassuring and that fetuses with preserved head growth may already have evolving abnormal brain development.13 Tolsa and colleagues14 evaluated longitudinal postnatal brain growth using MRI in 14 babies after FGR compared with gestation-matched controls. FGR was associated with reduced intracranial and cortical grey matter volumes in the first 2 weeks after birth, which persisted through to full term. These changes were associated with immature attention-interaction scores on behavioural testing.


Cerebral palsy

Because cerebral palsy (CP) is relatively uncommon (∼2.8 per 1000 live births15), trials and cohort studies are unlikely to show changes in CP rate relating to FGR, as most will be underpowered for this outcome. Evidence therefore comes from association studies in mainly epidemiological cohorts using birth weight as a surrogate for FGR. Indeed it has been estimated that reduced birth weight is causally associated with spastic CP: 22% of the attributable risk accruing from being below the 10th centile of the comparison population birthweight distribution.16

For some time it has been established that SGA babies at term are at higher risk than babies of appropriate weight for gestational age: in the National Collaborative Perinatal Project cohort born in 1959–66, the rates were 3.3 and 0.6 per 1000 children, respectively.17 The most recent data from a similarly large CP register collaboration endorse this: children born at 32–42 weeks of gestation had 4–6 times the prevalence of CP compared with children in the 25th–75th centile range,18 and the prevalence and severity of CP were higher at lower and higher z scores, respectively, of birth weight for gestation, in a reversed “J”-shaped distribution.18 19

The relationship between preterm birth, FGR and CP is much less clear, however. In the Western Australia Study, the risk of spastic CP associated with poor intrauterine growth appeared to depend on gestational age, with infants delivered at 34–37 weeks of gestation being at the highest risk (odds ratio of CP for children 34–37 weeks of gestation and <3rd centile at birth: 19.6; 95% CI 8.1 to 47), followed by those at term. In these carefully conducted studies, there was no association between FGR and CP at lower gestations.16 Other studies had also concluded that there was little increased risk of spastic CP in the very preterm SGA child (<32 weeks of gestation).2022 This phenomenon has also been seen in very-low-birthweight cohorts, where in some situations SGA initially appears to be protective against CP. However, such birthweight-based studies are confounded by the inclusion of many more children with FGR and greater maturity, and what appeared to be a significant reduction in risk (relative risk (RR): 0.1; 95% CI 0.03 to 0.6) disappeared when only babies <28 weeks of gestation were considered (RR: 1.2; 95% CI 0.2 to 6.4).23 It should be noted that the confidence intervals widen greatly, as the number of babies <28 weeks of gestation is considerably smaller than the whole cohort. Generally, it was considered that the effect of very preterm birth with its greatly increased risk of CP from perinatally acquired brain injuries may overwhelm the lesser association between FGR and CP.

However, further analysis of the Surveillance of Cerebral Palsy in Europe (SCPE) cohort suggests that the use of more appropriate fetal growth standards, which have less variance than neonatal birthweight reference data, reveals a similar association between birth weight for gestation and risk of CP over the whole gestational age range, so this may be in doubt.19 It will be interesting to re-evaluate these observations after the recently reported reduction in the frequency of bilateral spastic CP among infants of birth weight 1000–1499 g,24 many of which will have gestational ages <30 weeks; one might predict that the relationship between low birth weight for gestation and CP would become stronger.

Children with moderate and severe FGR who are born at near-term or term gestation therefore have a much increased risk of CP, whereas the same may not be true of very preterm infants. These data indicate that one of the fundamental problems in evaluating this literature is that babies born after 33 weeks of gestation are less likely to be delivered because of poor fetal health, and the association studies that work from a population of children with CP cannot impute aetiology. Studies of imaging results in term babies with spastic CP who were SGA indicates a range of findings that imply a variety of aetiologies.25 The causes of “CP” are poorly understood, and, of course, many come to light at some remove from the neonatal service and early diagnostic measures. Furthermore, early diagnosis of CP is relatively inaccurate,26 particularly in the ex-very-preterm child, in whom transient dystonia is common,27 and registers usually wait until at least 4 years of age before classifying the diagnosis. There is further potential for diagnosis bias because of disagreement between examiners, as some less severe cases may be relatively mild and variable.

Being SGA is the result of heterogeneous conditions, and current fetal medicine practice includes the investigation of the small, very preterm fetus and the use of termination where there are identified fetal abnormalities. The dichotomy in relationships seen may be simply because the aetiology of FGR is different in different populations and at lower gestations, that this process is relatively protective against causes underlying CP in less mature children, or because prematurity itself is so highly associated with CP secondary to haemorrhagic and ischaemic lesions that this effect overwhelms any effect of FGR, as suggested above. Given that the SGA baby who goes on to develop CP is often not identified before delivery, it is certainly possible that some developmental process or accident has occurred in utero, which leads to brain injury and also subsequent poor fetal growth. The CP may therefore not be caused by FGR per se but be part of the same causal pathway that leads to FGR and small size at birth.28

Neurodevelopment and cognition

Being born SGA is associated with lower intelligence, poor academic performance, low social competence and behavioural problems,2931 and many studies have shown that children born SGA (as variously defined) have long-term global cognitive impairments.3337

Early papers in general argue that there appears to be an inverse relationship between developmental scores and gestational age at birth, with a pregnancy of ∼36 weeks being a determinant for long-term prognosis. These effects seem to be enduring. For reviews, see Allen38 and Teberg et al.39

In the first series of longitudinal studies in relation to fetal growth by Harvey and colleagues,40 a cohort of SGA children born at full term, defined by longitudinal measures of fetal growth, were followed. The earlier in gestation that intrauterine head growth slowing was identified, the poorer was the outcome. Compared with children with normal head growth, SGA children with retardation of head growth before 26 weeks had deficits in perceptual performance, motor ability and general cognitive index, were less advanced developmentally,41 and had poorer school achievement with poor concentration compared with controls or those with later-onset growth slowing.42 Similar findings of reduced cognitive and motor scores have been reported in the longitudinal studies by Leitner and colleagues.4346 As part of the same study, compared with controls matched for prematurity and familial and socioeconomic factors, FGR was associated with deficits in short-term memory, but with preservation of recognition memory and super-span learning/memory and a normal rate of decay.33 The authors argue that their findings indicate an executive-attention deficit rather than hippocampal injury, which has been associated with prematurity and very low birth weight.47 No attempt has been made to relate these findings to fetal health measures.

The mainstay of assessment of fetal well-being is the fetal blood flow velocity pattern evaluated primarily in the umbilical artery. Decreasing diastolic flow velocities imply increasing placental vascular resistance, and, traditionally, absent or reduced end-diastolic flow velocities (AREDFVs) are used to guide delivery. Several studies have determined long-term outcome in relation to fetal flow velocities. In a series of studies in Lund, Sweden, children were followed through to 18 years of age, and outcomes were assessed in terms of whether AREDFV was present or not. At 6–7 years of age, there was an association between AREDFV and reduced cognitive scores,48 motor/minor neurological impairment49 and aortic growth/blood pressure,50 but not in terms of linear growth.51 A small subset of these children has been evaluated as young adults, aged 19 years. Impairment of cognitive function and vascular growth persisted.52 53 Using the optic fundus to evaluate neuronal development, poor retinal vascularisation was associated with increasing FGR,54 and there was found to be a loss of volume at the optic fundus thought to reflect reduced axonal size or number.55

In a study of outcome at 5–12 years in relation to fetal Doppler studies,56 reduced general cognitive scores and neurological abnormalities were observed in children with FGR and reversed end-diastolic flow velocity, but not in those with forward or absent flow velocities.

More recently, interest has focused on trying to evaluate regional blood flow in the fetus as a marker of placental perfusion and substrate supply. As substrate supply diminishes, the fetus redirects blood from non-essential to essential organs. This is manifested in the fetal cerebral circulation as increased fetal diastolic flow velocities as the vascular resistance decreases and flow increases.57 This is sometimes referred to as “brain sparing”, but is an attempt by the fetus to maintain substrate supply or to protect development.58 Indeed, as fetal hypoxia increases, further compensation is not possible and fetal vasoconstriction occurs. The obverse of this increased blood flow is a reduction in renal blood flow manifested in reduced fetal urine output and oligohydramnios.

These mechanisms are directed at fetal survival. It would seem logical that these processes indicate pre-existing hypoxia or substrate depletion and that the process of FGR is progressively associated with increasing risk of brain injury as the process evolves. Scherjon and colleagues59 have evaluated changes in cerebral blood flow velocity in relation to longer-term outcomes. In a longitudinal cohort study of 106 children with FGR born at 26–33 weeks’ gestation, accelerated visual maturation in infancy, using visual evoked cortical potentials, was observed compared with controls, and this was initially regarded as a beneficial adaptive process. This cohort was evaluated again at 3 years when developmental differences were attributable to cerebral circulatory changes, leading to the conclusion that the observation was a benign adaptive mechanism to protect the brain. However, after a further evaluation at 5 years, they reported that both the changes in cerebral Doppler and the acceleration of visual maturation were associated with a 9-point deficit in cognitive scores.60 Evaluation of these children in early infancy using these tests was not a good predictor of later cognitive outcome, which was impaired despite the adaptive changes observed.

What is already known on this topic

  • Being born small is a risk factor for cerebral palsy and for childhood learning difficulties, but the relationship between fetal growth and health is less clearly delineated.

What this review adds

  • Care should be taken not to confuse small size at birth with fetal growth restriction.

  • The relationship between cerebral palsy and small size at birth is robust and may be related to fetal growth restriction or to processes that lead to both fetal growth restriction and cerebral palsy.

  • Cognitive outcomes seem to be related to measures of fetal health, but management plans have not been shown to improve long-term outcomes as yet.

Children with poor longitudinal growth after FGR may respond well to growth hormone treatment.61 Recent interest has centred on the evaluation of growth hormone as a mediator of, or potential remediation for, these cognitive deficits. After 2 years of treatment with human growth hormone, improved IQ, social acceptance and self-worth were observed.62 After 8 years of treatment, improved IQ, behaviour and self-perception were observed, with scores moving from below average to approaching Dutch norms.63 The authors argue that treatment with human growth hormone has a direct effect on cerebral functioning, particularly performance of the right hemisphere. Cognitive improvement was independent of height, although behavioural problems improved in correlation with height—that is, the nearer to peer height reached, the fewer the behavioural problems.

It seems clear that randomised studies must determine the extent to which decision points concerning delivery may alter these long-term outcomes. To date, the only study to do this is GRIT. In this, children were randomly assigned to immediate or deferred delivery when the correct management was uncertain. At 2 years of age, the outcome, assessed using the Griffiths’ Scales of Infant Development, showed no differences between those in each group.7 As we have seen from other studies,60 this may be too early to detect important changes in childhood outcomes for this group. We have just completed an assessment of the GRIT cohort in middle childhood, and the results will be available later in 2008. Meanwhile a further randomised trial of delivery based on changes in the ductus venosus waveform is underway and will help to refine delivery criteria further (TRUFFLE Study;


Measures of fetal well-being made before birth may have important consequences for long-term neurodevelopmental outcomes, and there is good evidence that brain development is already affected in the first 2 weeks after birth. It is less clear whether the same processes lead to both focal haemorrhagic and ischaemic lesions (with increased risk of CP) and the generalised brain injury probably associated with poorer cognitive outcomes, as these studies have not been carried out. Long-term impairment in cognitive, executive, motor and behavioural functions have all been described, and they appear to persist through to adult life. These impairments lie in conjunction with the real concerns about long-term cardiovascular health following poor intrauterine growth, which are outside the remit of this review. In turn, this sits alongside the anxiety that poor fetal health may also act as a trigger for other diseases such as schizophrenia.64 Generally, however, these childhood impairments do not appear to produce worse outcomes in terms of self-reported health-related quality of life.65


D-MW was supported by the Medical Research Council.


View Abstract


  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.