Article Text

Neonatal EEG and neurodevelopmental outcome in preterm infants born before 32 weeks
  1. Maximilien Périvier1,
  2. Jean-Christophe Rozé1,2,3,
  3. Géraldine Gascoin2,4,
  4. Matthieu Hanf3,
  5. Bernard Branger2,
  6. Valérie Rouger2,3,
  7. Isabelle Berlie2,5,
  8. Yannis Montcho2,6,
  9. Yann Péréon7,
  10. Cyril Flamant1,2,3,
  11. Sylvie Nguyen The Tich2,5
  1. 1Department of Neonatal Medicine, University Hospital of Nantes, Nantes University, Nantes, France
  2. 2‘Loire Infant Follow-up Team’ (LIFT) Network, Nantes, Pays de Loire, France
  3. 3Clinical Research Center, INSERM CIC004, University Hospital of Nantes, Nantes, France
  4. 4Department of Neonatal Medicine, Angers University, University Hospital of Angers, Angers, France
  5. 5Department of Pediatric Neurology, Angers University, University Hospital of Angers, Angers, France
  6. 6Department of Neonatal Medicine, Hospital of Le Mans, Le Mans, France
  7. 7Laboratoire d'Explorations Fonctionnelles, Nantes University, Centre de Référence Maladies Neuromusculaires Nantes-Angers, University Hospital of Nantes, Nantes, France
  1. Correspondence to Professor Sylvie Nguyen The Tich, Department of Pediatric Neurology, Angers University, University Hospital of Angers, bâtiment Robert Debré, 4 rue Larrey, ANGERS, 49000, France; sylvie.nguyenthetich{at}chru-lille.fr

Abstract

Objective To assess the value of neonatal EEG for predicting non-optimal neurodevelopmental outcomes in very preterm infants, using a multimodal strategy of evaluation comprising brain imaging and clinical assessment.

Design and setting Between 2003 and 2009, we performed an observational, population-based study. Out of 2040 eligible preterm infants born before 32 weeks, 1954 were enrolled in the French regional Loire Infant Follow-Up Team (LIFT) cohort. 1744 (89%) of these completed the follow-up. Neonatal EEGs were recorded prospectively as two EEGs during the first 2 weeks of life and then one every 2 weeks up to 33 weeks.

Main outcome measures The neurodevelopmental outcome was assessed by physical examination, the Brunet–Lézine Test and/or the Age and Stages Questionnaire at 2 years of corrected age.

Results Of the 1744 infants assessed at 2 years, 422 had a non-optimal outcome. A total of 4804 EEGs were performed, and 1345 infants had at least one EEG. EEG abnormalities were predictive of non-optimal outcomes after controlling for confounding factors such as severe intracranial lesions detected by brain imaging. Transient moderate and severe abnormalities were independent predictors of non-optimal outcomes with an OR and 95% CI of 1.49 (1.08 to 2.04) and 2.38 (1.49 to 3.81), respectively. In the validation group, the predictive risk stratification tree identified severe abnormalities as a factor contributing to the prognosis of two subgroups: infants with severe cranial lesions and infants with a normal examination at discharge and without severe cranial lesions.

  • Neonatology
  • Neurodevelopment
  • Clin Neurophysiology

Statistics from Altmetric.com

What is already known on this topic

  • Neurodevelopmental outcome is one of the most important concerns with very preterm children. Based on small population studies, abnormal conventional neonatal EEGs have been associated with poor neurodevelopmental outcomes.

What this study adds

  • This large population-based study is the first one to demonstrate the usefulness of neonatal EEGs surveillance included in a multimodal strategy for predicting neurodevelopmental outcome.

  • The more severe the EEG surveillance, the more the neurodevelopmental outcome was poor and the more the association was tight.

  • We propose a multimodal evaluation, including EEG surveillance, clinical assessment at discharge and cerebral imaging, to increase neonatal screening capacity for a more efficient follow-up.

Introduction

Despite improvements in treatment and survival, very preterm infants, that is, those born before 32 weeks, have a high risk for neurodevelopmental disabilities.1 Predicting neurodevelopmental outcome is a challenge for physicians, and it is also a pronounced parental concern. Early recognition of at-risk infants may allow for early intervention to improve the outcome. Since systematic surveillance programmes are hard to sustain on a large scale, it is useful to identify low-risk and high-risk infants in order to provide a more efficient follow-up.2 ,3

Cerebral imaging, neurophysiological methods and clinical examination are used in neonatal intensive care units (NICUs), but none is sufficient by itself.4–6

Neonatal conventional EEG is a technique widely used in France. Two recent studies7 ,8 have confirmed its previously recognised predictive value.9–13 The aim of this study was to evaluate the predictive value of neonatal EEG by itself and when included in a multimodal evaluation comprising EEG surveillance, clinical assessment at discharge and cerebral imaging in terms of neurodevelopmental status at 2 years of corrected age in very preterm infants in a large, prospective, population-based regional cohort study.

Methods

Study design

We performed an observational, population-based, regional cohort study that included infants born before 32 weeks in the French region ‘Pays de La Loire’ from 1 March 2003 to 31 December 2009 (Loire Infant Follow-Up Team (LIFT) cohort). Preterm infants were hospitalised in nine NICUs, two of which are university hospitals. They were all meant to apply the same surveillance protocol, although some deviations from this could occur depending on local resources. The cohort was registered with the French National Ethics Committee (Commission Nationale de l'Informatique et des Libertés n°851117) in order to collect clinical data. Parents provided written informed consent before each inclusion in the LIFT cohort and before use of the data. Specific approval to use the data in this study was obtained from the Institutional Review Board of the University Hospital of Nantes.

The following data were collected: gestational age, birth weight Z-score,14 sex, neonatal course, neurological assessment at discharge, brain imaging and EEG results.

The neurological assessment at discharge was performed by a trained neonatologist using the Amiel–Tison method and was classified into four categories from normal to severely abnormal as previously published.4 The brain imaging protocol comprised sequential cranial ultrasound (CUs): at least one during the first week of life, then every 2 weeks until the fourth week of life or 32 weeks of postmenstrual age (PMA). A brain MRI at term-equivalent age was not performed systematically. Rather, it was done to identify the nature of the cerebral lesions picked up by CUs.

The results from brain imaging were ranked into two categories: the absence of a severe cerebral lesion (normal CUs or intraventricular haemorrhage (IVH) Grade I or II)15 or the presence of a severe cerebral lesion (IVH III, white matter disease (periventricular leucomalacia, ventriculomegaly), IVH and white matter disease). In case of more than one lesion, the worst one was retained.

The EEG surveillance was established in accordance with the French guidelines:16 ,17 one EEG in the first week of life, one in the second week and then one every 2 weeks up to 33 weeks of PMA. EEGs were performed with cup electrodes using reduced montages with eight (FP1, FP2 C3 C4 O1 O2 T3 T4) or 11 (adding FZ CZ PZ) electrodes depending on the head size, heart and respiratory rates. The clinical status (stable or unstable) and the administration of sedative drugs were noted by the technicians present during the entire recording. The minimal duration for recording was 45 min. The EEG was analysed by six trained neurophysiologists using the French glossary,17 who were in close collaboration with each other and who were supervised by one of the authors (SNTT). Each EEG was ranked as normal (normal background activity for the corrected age and no abnormal features), moderately abnormal (increased discontinuity with maximal interburst interval duration less than one and a half the maximal value for the age group, abnormal delta brushes, dysmature patterns, positive temporal sharp waves and positive rolandic sharp waves less than 1/min, disorganised patterns) or severely abnormal (excessive discontinuity with maximal interburst interval duration above one and a half the maximal value for the corrected age, seizures, the absence of sleep cycles and positive rolandic sharp waves at more than 1/min). If the EEG was considered to be artifactual or modified by sedative drugs, a supplementary EEG could be ordered.18 Illustrative EEGs are provided in online supplementary figure S1. For each infant, the entire EEG surveillance was scored based on the degree of abnormalities (none, moderate or severe) and their persistence (transient (one or two abnormal EEGs) or persistent (more than two abnormal EEGs)) in three grades: normal (Grade 0), transient moderately abnormal EEGs (Grade 1) and very abnormal EEGs with persistent or severe abnormalities (Grade 2).

Neurodevelopmental outcome

Neurological assessment was performed by a trained paediatrician. It was deemed to be non-optimal if there was no ability to walk independently at 2 years of age, or when gait abnormalities or any abnormal clinical examinations during independent walking were noted. Psychomotor evaluation was performed by a specialised psychologist using the revised Brunet–Lézine Test; a French developmental test providing a developmental quotient (DQ).19 When the psychological evaluation could not be performed, we used the Ages & Stages Questionnaire (ASQ); a parent-completed developmental assessment tool validated by comparison with the revised Brunet–Lézine Test.20 Psychomotor development was considered non-optimal when the child could not take the Brunet–Lézine Test because of neurological impairment, when the Brunet–Lézine DQ was below 85 or when the ASQ score was ≤185. Lastly, infants with either non-optimal neuromotor function or abnormal psychomotor development were included in the non-optimal neurodevelopmental group.21

Statistical methods

The neonatal data for assessed and non-assessed infants were compared using the χ2 test or Fisher's exact test, when necessary. The association between the EEG grades and neurodevelopmental outcomes was performed in an unadjusted analysis among assessed children. The neurodevelopmental outcome at 2 years of age was treated as a binary variable: optimal or non-optimal. The sensitivity, specificity and positive and negative likelihood ratios were determined for each EEG grade using the Diagnostic Test Calculator (Department of Medical Education, University of Illinois, Chicago, Illinois, USA: http://araw.mede.uic.edu/cgi-bin/testcalc.pl). The association between each EEG grade and the outcome at 2 years was determined using a logistic regression with and without adjustment for gestational age, birth weight Z-score, sex, neurological assessment at discharge and cerebral lesions. The same association was determined among different subgroups (ie, infants with EEG, with imaging, with MRI, with EEG and imaging, with or without DQ, hospitalised in university or non-university NICUs) in eight sensitivity analyses. Lastly, a risk stratification tree for non-optimal developmental outcomes at 2 years of corrected age was built into a training set and validated into a validation set. The grade of cerebral lesions, rank of the neurological assessment at discharge and EEG grade were included in the model and prioritised using a Chi-squared automatic interaction detector analysis (CHAID).22 At each step, CHAID chose the independent variable (neurological examination at term, EEG grade, severe cerebral lesions) that had the strongest interaction with the outcome status at 2 years. This variable became the first branch in a tree with a leaf for each category significantly linked to a different outcome. The CHAID analysis was run with parent nodes defined at 40 subjects, child node defined at 10 subjects and significance set at ≤0.05. The validation data set was used to confirm the items’ prioritisation and diagnostic trees. The significance level was set at 0.05 for each two-tailed comparison. All the analyses were performed with SPSS V.17.0 (SPSS Inc., Chicago, Illinois, USA).

Results

At 2 years of corrected age, 1744 children out of the 1954 enrolled in the LIFT cohort were assessed (figure 1): 1612 (92.4%) by neurological examination and 132 using the questionnaire. Nine hundred and sixty-eight children underwent a Brunet–Lézine Test, while 1540 had an ASQ and 926 had both. Assessed children were more immature (table 1). The neurodevelopmental outcome was considered as non-optimal for 422 out of the 1744 children (24.2%). The neurological examination was abnormal for 184 children, including 14 with no standing position, 92 with abnormal gait and 78 with a normal gait but an abnormal clinical examination. The DQ was below 85 for 101 children. The ASQ score was below 185 for 137 children.

Table 1

Comparison between assessed and non-assessed infants at 2 years of corrected age

Figure 1

Flow chart. LIFT, Loire Infant Follow-Up Team.

During the neonatal period, cerebral imaging by CUs was performed on 1598 infants (91.6%) and by MRI on 292 infants (16.7%). The median (IQR) number of CUs was 4 (2–5), 2 (2–3) and 2 (1–3) for infants born before 28 weeks, between 28 and 29 weeks and between 30 and 31 weeks, respectively. For 132 infants (7.6%), the result was deemed to be a ‘severe cerebral lesion’ (table 1).

Neonatal EEG monitoring (4804 EEGs) was performed with 1345 infants (77.1%). EEGs were performed more frequently in university NICUs (88.2 vs 29.4%, p<0.001). The median (IQR) number of EEGs was 4 (2–5), 3 (2–4) and 2 (1–3) for infants born before 28 weeks, between 28 and 29 weeks and between 30 and 31 weeks, respectively. Neurodevelopmental outcome did not significantly differ between preterm infants with no EEG and with a normal EEG surveillance. EEG grades were associated with non-optimal outcome (figure 2). The sensitivity, specificity, positive and negative likelihood ratios and their 95% CIs of Grade 2 EEGs were 0.16 (0.13 to 0.19), 0.95 (0.94 to 0.96), 3.41 (2.50 to 4.67) and 0.88 (0.85 to 0.92), respectively. The positive likelihood ratio of the combination of Grade 2 EEG and severe cerebral lesions was 19.8 (9.0 to 43). Detailed sensitivity, specificity, positive and negative likelihood ratios and 95% CIs of neurological assessment, imaging, EEG and their combinations are indicated in online supplementary table. Among the 1345 infants with EEG and neurodevelopmental assessment at 2 years, the adjusted OR (aOR) for non-optimal neurodevelopmental outcome was 2.4, 95% CI (1.5 to 3.8). Crude OR and aOR are provided for each EEG grade in table 2, and in online supplementary figure S2 for Grade 2 EEG among different subgroups of preterm infants.

Table 2

EEG surveillance abnormalities as a risk factor for global neurodevelopment at 2 years of corrected age analysis

Figure 2

The rate of non-optimal neurodevelopmental outcome at 2 years according to the neonatal EEG surveillance abnormalities: all normal (Grade 0), transient moderate abnormalities (Grade 1) and permanent moderate abnormalities or severe abnormalities (Grade 2). Circles indicate the percentage of infants with a non-optimal outcome. Intervals indicate the 95% CI.

A predictive risk stratification tree (figure 3) identified Grade 2 EEG as a predictor contributing to the prognosis in two subgroups: infants with severe cranial lesions and infants with normal examination at discharge and no severe cranial lesions.

Figure 3

Classification and regression tree predicting non-optimal neurodevelopmental outcomes in the validation group. Each node shows the selected splitting variable, number and 95% CI of infants with a non-optimal outcome. The terminal nodes marked in grey represent the subgroups of infants among which Grade 2 EEG surveillance abnormalities increased the risk. IVH, intraventricular haemorrhage; PVL, periventricular leukomalacia.

Discussion

In this population-based study, we have demonstrated that our grading of neonatal EEG surveillance was significantly associated with neurodevelopmental outcomes at 2 years of corrected age. Thus, we found that the worse the EEG abnormalities, the higher the rate of non-optimal developmental outcome and the stronger the association. This is the first study to establish the predictive value of neonatal EEGs on this scale. A risk stratification tree identified Grade 2 EEG as a factor contributing to the prognosis of two subgroups: infants with severe cranial lesions and infants with normal examination at discharge and without severe cranial lesions.

Cerebral imaging is the most widely used approach during neonatal evaluation. Abnormalities in late CUs and near-term MRI were found to be associated with compromised outcomes.23 CUs provides a good screening tool to detect serious brain injuries that result in motor handicaps,24 although its ability to accurately diagnose non-cystic lesions or predict isolated cognitive impairment is limited.5 ,25 ,26 MRI at term-equivalent age had been shown to have a high predictive value for psychomotor impairment at 2 years of age, and it can be used to stratify these infants according to risk.23 ,24 ,27 Analysis of moderate white matter signal abnormalities is still a contentious issue.24 ,28 The use of MRI in daily practice remains limited by its cost, its accessibility and the expertise that it requires.5 ,24

The historical value of EEGs arose from early studies conducted when the mortality rate and the incidence of severe brain injury were high among preterm infants, and who were hence not always fully assessed in terms of their cognitive or psychological outcomes.9 ,11 ,12 Recent studies have confirmed the relevance of neonatal EEG and have refined some patterns.7 ,8 We propose a new original approach to EEG results based on grading of the entire EEG surveillance to assess the neonatal cerebral damage. We believe that it provides an adequate assessment of preterm cerebral development. Sequential repeated EEGs, each representing a ‘snapshot’ of the brain's functioning, can detect transient or more persistent abnormalities. This neurophysiological monitoring under various clinical situations provides biomarkers of the complex amalgam of destructive and developmental mechanisms that characterise the encephalopathy of prematurity.29

The specificity of EEGs was good, but their sensitivity was low. Combining EEGs and imaging dramatically increased the positive likelihood ratio, highlighting the usefulness of a multimodal approach. With our risk stratification tree, we identified a low-risk group for whom prematurity seemed to be a transient birth event, and a high-risk group for whom prematurity opened the door of a chronic condition.29 Follow-up programmes should rely on optimised neonatal screening strategies. Since most very preterm children exhibited a good neurodevelopmental outcome, a normal neonatal multimodal evaluation could help to reassure families. However, a certain degree of follow-up remains necessary as nearly 20% of the infants had a non-optimal outcome at 2 years despite a completely normal neurological assessment in the NICU.

One limitation of this cohort study was that the evaluation at 2 years was performed by more than 120 paediatricians. Although this may add a degree of subjectivity, despite the use of a standardised clinical examination, it reflects a real-life situation. Moreover, the assessment at 2 years of age may underestimate the child's cognitive, subtle motor or behavioural abnormalities. The children were monitored until 5 years of age, and further analyses are now taking place to investigate these possibilities. Other limitations were the small number of MRIs that were taken, the lack of quantitative EEGs, the number of EEGs performed not being strictly the same for each child and the number of infants receiving EEGs varying according to the type of NICU. EEGs were performed less frequently in non-university NICUs due to a lesser availability. Thus, a bias was observed: preterm infants hospitalised in these NICUs and who had a Grade 2 EEG were at higher risk of non-optimal development (see online supplementary figure S2). Nevertheless, the association between Grade 2 EEG and non-optimal outcome was strong and consistent in all eight sensitivity analyses.

Performing EEGs in preterm infants is still considered as a highly specialised and skilled technique. Recent neonatal EEG and aEEG current guidelines have refined normal and abnormal aspects.7 ,8 ,10–13 ,16 ,17 The French glossary17 has been supplemented with the American guidelines on neonatal EEG, including the preterm aspects.30 Simplifying the interpretation of neonatal EEG is possible, using quantitative method of signal analysis framing the amplitude (aEEG) and other features like spectral content or specific physiological patterns like delta brushes31–33 that are useful for long-term monitoring. Even though the predictive value of aEEG has mainly been studied among term neonates with hypoxic–ischaemic encephalopathy or seizures,34–36 these methods seem to have ample potential and should be developed for preterm applications to allow more ready use of EEG and to stimulate wider use of brain monitoring techniques.30 ,32 ,37

For repeated EEGs, we recommend one EEG shortly after birth, since signs of birth asphyxia are clinically silent in preterm babies and may disappear rapidly.17 This first EEG may also detect the cerebral consequences of some antenatal conditions. Surveillance should then be performed by scheduled EEGs every 2 weeks and stopped after three normal EEGs or after 33 weeks.10 ,38 ,39 Supplementary EEGs should be ordered if at-risk situation for the brain occurs (eg, haemodynamic failure and septic shock).

Conclusion

Neonatal EEG surveillance exhibited a good specificity and a good positive likelihood ratio for neurodevelopmental outcomes in very preterm infants assessed at 2 years of corrected age. By adding EEG in a multimodal predictive evaluation, the neonatal screening capacity is increased, thereby providing a more efficient follow-up and stronger family support.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors MP: drafted the initial manuscript, carried out the initial analyses and approved the final manuscript as submitted. J-CR: conceptualised and designed the study, carried out the initial analyses, reviewed and revised the manuscript and approved the final manuscript as submitted. GG, YM: coordinated and supervised data collection at one site, reviewed and revised the manuscript and approved the final manuscript as submitted. MH, BB: coordinated and supervised data collection at all sites, carried out the initial analyses and approved the final manuscript as submitted. VR, IB and YP: coordinated and supervised data collection at all sites, reviewed and revised the manuscript and approved the final manuscript as submitted. CF: carried out the initial analyses, reviewed and revised the manuscript and approved the final manuscript as submitted. SNTT: conceptualised and designed the study, carried out the initial analyses, drafted the initial manuscript and approved the final manuscript as submitted.

  • Funding The Loire Infant Follow-Up Team (LIFT) cohort was supported by grants from the Regional Health Agency of Pays de la Loire.

  • Competing interests None declared.

  • Ethics approval CNIL, France.

  • Provenance and peer review Not commissioned; externally peer reviewed.

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.