Intended for healthcare professionals

CCBYNC Open access

Neurodevelopmental outcomes at age 5 among children born preterm: EPIPAGE-2 cohort study

BMJ 2021; 373 doi: (Published 28 April 2021) Cite this as: BMJ 2021;373:n741

Read our latest coverage of the coronavirus pandemic

  1. Véronique Pierrat, neonatologist1 2,
  2. Laetitia Marchand-Martin, statistician1,
  3. Stéphane Marret, professor in paediatrics3 4,
  4. Catherine Arnaud, epidemiologist5 6 7,
  5. Valérie Benhammou, health research engineer1,
  6. Gilles Cambonie, professor in paediatrics8,
  7. Thierry Debillon, professor in paediatrics9 10,
  8. Marie-Noëlle Dufourg, epidemiologist11,
  9. Catherine Gire, neonatologist12,
  10. François Goffinet, professor in obstetrics and gynaecology1 13,
  11. Monique Kaminski, epidemiologist1,
  12. Alexandre Lapillonne, professor in paediatrics14,
  13. Andrei Scott Morgan, neonatologist and epidemiologist1 15,
  14. Jean-Christophe Rozé, professor in paediatrics16 17,
  15. Sabrina Twilhaar, research psychologist1,
  16. Marie-Aline Charles, professor in public health and epidemiology9 18,
  17. Pierre-Yves Ancel, professor in public health and epidemiology1 19
  18. on behalf of the EPIPAGE-2 writing group
    1. 1University of Paris, CRESS, Obstetrical Perinatal and Pediatric Epidemiology Research Team, EPOPé, INSERM, INRAE, F-75004 Paris, France
    2. 2CHU Lille, Department of Neonatal Medicine, Jeanne de Flandre Hospital, Lille, France
    3. 3Department of Neonatal Medicine, Intensive Care, and Neuropediatrics, Rouen University Hospital, Rouen, France
    4. 4INSERM U1254, Neovascular Team, Perinatal Handicap, Institute of Biomedical Research and Innovation, Normandy University, Rouen, France
    5. 5INSERM U1027, SPHERE Team, France
    6. 6Toulouse 3 University Paul-Sabatier, Toulouse, France
    7. 7Clinical Epidemiology Unit, University Hospital Toulouse, Toulouse, France
    8. 8Department of Neonatal Medicine, Montpellier University Hospital, Montpellier, France
    9. 9Neonatal Intensive Care Unit, Grenoble Alps University Hospital, Grenoble, France
    10. 10UMR 5525 Techniques pour l'Evaluation et la Modélisation des Actions de la Santé, Centre National de la Recherche Scientifique, Techniques de l'Ingénierie Médicale et de la Complexité-Informatique, Mathématiques et Applications, Grenoble Alps University, Grenoble, France
    11. 11French Institute for Demographic Studies, French Institute for Medical Research and Health, French Blood Agency, ELFE Joint Unit, Paris, France
    12. 12Department of Neonatology, North Hospital, Assistance Publique des Hôpitaux de Marseille, Marseille, France
    13. 13Port-Royal Maternity, Assistance Publique-Hôpitaux de Paris, Centre-University of Paris, Federation Hospitalo-Universitairefor prematurity, Paris, France
    14. 14Assistance Publique-Hôpitaux de Paris, University Hospital Necker-Enfants Malades, University of Paris, Paris, France
    15. 15UCL Elizabeth Garrett Anderson Institute for Women's Health, University College London, London, UK
    16. 16Department of Neonatal Medicine, Nantes University Hospital, Nantes, France
    17. 17Clinical Epidemiology, Clinical Investigation Centre CIC004, Nantes University Hospital, Nantes, France
    18. 18University of Paris, CRESS, Team Early Life Research on Later Health, UMR 1153, INSERM, INRAE, Villejuif, France
    19. 19Clinical Investigation Centre P1419, Assistance Publique-Hôpitaux de Paris, Paris, France
    1. Correspondence to: V Pierrat, Tenon Hospital, Equipe EPOPé, U1153, 4 Rue de la Chine, 75020 Paris, France veronique.pierrat{at}
    • Accepted 4 March 2021


    Objectives To describe neurodevelopment at age 5 among children born preterm.

    Design Population based cohort study, EPIPAGE-2.

    Setting France, 2011.

    Participants 4441 children aged 5½ born at 24-26, 27-31, and 32-34 weeks

    Main outcome measures Severe/moderate neurodevelopmental disabilities, defined as severe/moderate cerebral palsy (Gross Motor Function Classification System (GMFCS) ≥2), or unilateral or bilateral blindness or deafness, or full scale intelligence quotient less than minus two standard deviations (Wechsler Preschool and Primary Scale of Intelligence, 4th edition). Mild neurodevelopmental disabilities, defined as mild cerebral palsy (GMFCS-1), or visual disability ≥3.2/10 and <5/10, or hearing loss <40 dB, or full scale intelligence quotient (minus two to minus one standard deviation) or developmental coordination disorders (Movement Assessment Battery for Children, 2nd edition, total score less than or equal to the fifth centile), or behavioural difficulties (strengths and difficulties questionnaire, total score greater than or equal to the 90th centile), school assistance (mainstream class with support or special school), complex developmental interventions, and parents’ concerns about development. The distributions of the scores in contemporary term born children were used as reference. Results are given after multiple imputation as percentages of outcome measures with exact binomial 95% confidence intervals.

    Results Among 4441 participants, 3083 (69.4%) children were assessed. Rates of severe/moderate neurodevelopmental disabilities were 28% (95% confidence interval 23.4% to 32.2%), 19% (16.8% to 20.7%), and 12% (9.2% to 14.0%) and of mild disabilities were 38.5% (33.7% to 43.4%), 36% (33.4% to 38.1%), and 34% (30.2% to 37.4%) at 24-26, 27-31, and 32-34 weeks, respectively. Assistance at school was used by 27% (22.9% to 31.7%), 14% (12.1% to 15.9%), and 7% (4.4% to 9.0%) of children at 24-26, 27-31, and 32-34 weeks, respectively. About half of the children born at 24-26 weeks (52% (46.4% to 57.3%)) received at least one developmental intervention which decreased to 26% (21.8% to 29.4%) for those born at 32-34 weeks. Behaviour was the concern most commonly reported by parents. Rates of neurodevelopment disabilities increased as gestational age decreased and were higher in families with low socioeconomic status.

    Conclusions In this large cohort of children born preterm, rates of severe/moderate neurodevelopmental disabilities remained high in each gestational age group. Proportions of children receiving school assistance or complex developmental interventions might have a significant impact on educational and health organisations. Parental concerns about behaviour warrant attention.


    Children born preterm may experience many developmental difficulties, varying from severe to subtle impairments.12 Most of these deficits are better identified after age 2, when higher order cognitive, language, and behavioural processes develop and affect early sensorimotor abilities. Neurodevelopmental disabilities in early childhood are commonly reported as a composite outcome of cerebral palsy, sensory and cognitive impairments.3456 More subtle deficits, such as behavioural difficulties or developmental coordination disorders, are usually reported separately,78910 even though they may alter a child’s socioemotional competencies, affect daily activities and school outcomes, disturb family functioning, and require substantial medical, social, educational, and family support.11 Thus it is important that all conditions are considered together. Additionally, the parents’ viewpoint is increasingly recognised as essential to help bridge gaps between the information parents desire on outcomes and that which has traditionally been offered to them.1213

    Advances in perinatal intensive care, including increased use of antenatal corticosteroids, of exogenous surfactant, of non-invasive respiratory support, and avoidance of the liberal use of postnatal corticosteroids to treat bronchopulmonary dysplasia,14 have led to increased survival of infants born preterm and particularly those born extremely preterm (24-26 weeks).1516 Recent large population based studies reporting outcomes in early childhood have mainly focused on those born extremely preterm.35617 Long term outcomes in this group have not convincingly improved.14 Children born very (27-31 weeks) or moderately (32-34) preterm have generated less research,18 but changes in care have also been substantial for these infants, who represent larger proportions of those facing developmental difficulties. Up to date data are thus essential for evaluating their needs and providing stakeholders with useful information.

    The EPIPAGE-2 cohort is a national cohort designed to investigate outcomes for children born below 35 weeks in 2011 in France.19 Improvements in survival, neonatal outcomes, and neurosensory disabilities at age 2 have already been shown between the 1997 EPIPAGE cohort and EPIPAGE-2.1620 We then conducted an assessment at age 5½. The aim of the present study was to describe neurodevelopmental outcomes, including cerebral palsy, sensory impairments, and cognition, as well as behavioural difficulties and developmental coordination disorders, for children included in the EPIPAGE-2 follow-up using contemporary children born at term as a reference. To further assess the family and social burden of prematurity, we also report educational assistance, use of complex developmental interventions,21 and parental concerns about development.


    Study design

    EPIPAGE-2 is a French prospective national population based cohort of preterm born children. All births at 22 to 34 weeks’ gestation in all maternity units in 25 French regions (21 of the 22 metropolitan regions and all overseas regions) were eligible for inclusion. The study began on 28 March 2011. We recruited infants born at 22-26 weeks during an eight month period, those born at 27-31 weeks during a six month period, and those born at 32-34 weeks, during a five week period.19 All survivors were invited to participate in a comprehensive neurodevelopmental assessment at age 5½. To obtain reference data for the various developmental tests used in this study and describe outcomes for the 5½ year old EPIPAGE-2 cohort, we aimed to assess term born children from the ELFE (Étude Longitudinale Française depuis l’Enfance) cohort with the EPIPAGE-2 follow-up protocol. The ELFE cohort is a contemporary French cohort of more than 18 000 children born after 32 weeks in 2011, in 344 randomly selected public and private maternity units in metropolitan France.22 For financial and organisational reasons, the assessment could be offered only to 600 children born at term from the ELFE cohort. The size of this reference group was sufficient to obtain good precision for estimated scores of the main scales used based on known means and standard deviations (web appendix 1). Consent for participation was obtained from parents of 592 of the 600 selected term born children.

    Data collection

    Assessments included an interview with parents, a self-administered parental questionnaire, clinical examination by a paediatrician, and an evaluation by a psychologist, performed in 110 centres specifically opened for the study. Paediatricians and psychologists were trained to ensure homogeneity in their evaluations. Assessors were not blinded to the child’s gestational age. If the child was too disabled to participate, parental permission was sought to contact rehabilitation centres to provide information enabling classification of disabilities. In one region (accounting for 13 subjects), and elsewhere for parents refusing formal assessment or if the assessment team was not available, parents were asked to complete a postal questionnaire including information on cerebral palsy, severe/moderate sensory disabilities, school attendance, and use of complex developmental interventions.

    Neurodevelopmental evaluation

    Description and classification of variables used to describe neurodevelopment and classification of neurodevelopmental disabilities are described in table 1.

    Table 1

    Definition of neurodevelopmental outcome measures at 5½ years and classification of neurodevelopmental disabilities

    View this table:

    Motor domain—motor function was evaluated during the clinical examination, designed to screen cerebral palsy and developmental coordination disorders. Cerebral palsy was reported according to the diagnostic criteria of the Surveillance of Cerebral Palsy in Europe network,23 and graded using the Gross Motor Function Classification System (GMFCS).24 Developmental coordination disorders were defined as a total score less than or equal to the fifth centile of the reference sample on the Movement Assessment Battery for Children, second edition,25 in those children of the reference sample who did not have cerebral palsy, severe/moderate sensory disability, or a full scale intelligence quotient score less than two standard deviations below the mean.

    Sensory domain—we based assessments of hearing and vision on interviews with parents and medical records available during the clinical examination. Visual acuity was evaluated with the Sander-Zalonghi scale.26 Degree of visual disability was classified according to World Health Organization criteria,27 and hearing disability according to the severity of hearing loss and the need for hearing aids.28

    Cognitive domain—we used the French version of the Wechsler Preschool and Primary Scale of Intelligence, fourth edition,29 which measures cognitive ability in five domains (verbal comprehension, visuospatial indices, fluid reasoning, working memory, processing speed), resulting in a composite full scale intelligence quotient reflecting the overall level of intelligence. Results are reported as mean and standard deviations, and by categories using cut-off points according to the mean and standard deviations of the reference sample.

    Behaviour—we used the strengths and difficulties questionnaire,30 a self-administered parental questionnaire designed to screen for symptoms of hyperactivity inattention, emotional, conduct, and peer problems. A total score is calculated by summing the scores of each domain, with a higher score indicating more difficulties. The questionnaire is not used for diagnosis. Children with a total score greater than or equal to the 90th centile of a reference sample, however, are usually considered as having a substantially raised probability of formally diagnosed mental health problems.31

    Classification of neurodevelopmental disabilities

    Information on cerebral palsy, sensory deficits, cognitive, behavioural, and motor coordination functioning were combined into a composite measure with four levels of neurodevelopmental disabilities (table 1). Severe and moderate disabilities were primarily based on the severity of cerebral palsy, sensory impairment, and cognitive deficiency, as their presence rules out a diagnosis of developmental coordination disorders, and comorbid behavioural difficulties are highly prevalent. Both developmental coordination disorders and behavioural difficulties were included in the classification of mild neurodevelopmental disabilities.

    Education, complex developmental interventions, and parental concerns

    Parents were asked about education, use of complex developmental interventions, and concerns about their child’s development to better understand the effect of developmental outcomes on daily family functioning. Variables collected, recorded as binary (yes/no) unless specified, were school enrolment and, if yes, type of school (mainstream class without support, mainstream class with support—defined as either part or full time face to face learning support provided by a dedicated professional—or special schooling—specifically adapted for disabled children); use of complex developmental interventions (at least two visits during the preceding 12 months to a psychologist or psychiatrist, orthoptist, speech therapist, occupational therapist, or physiotherapist) or care in a rehabilitation centre; a request for assistance made to a Maison Départementale des Personnes Handicapées, a French institution that provides additional support for disabled children (specialised health or educational services) or financial support for parents. Finally, parents indicated whether or not they had concerns about their child’s language, coordination of movements, learning, or behaviour.

    Data management and statistics

    Results among survivors at 5½ years are presented according to groups of gestational age at birth (24-26, 27-31, and 32-34 weeks). Firstly, summary data on child (sex, small for gestational age32) and maternal characteristics at birth (age, country of birth, cohabitation status, level of education, and parental socioeconomic status), obstetric factors (primiparous, multiple pregnancy, antenatal steroids, caesarean section), the neonatal period (severe neonatal morbidities, including severe bronchopulmonary dysplasia, necrotising enterocolitis (Bell stages 2-3), retinopathy of prematurity stage >3, or any intraventricular haemorrhage grade III-IV or cystic periventricular leukomalacia on cranial ultrasonography16), and outcome at 2 years (cerebral palsy, a score below threshold in at least one domain of the Ages and Stages Questionnaire33) were compared between children seen and not seen at follow-up. Those children having formal assessments were also compared with those who completed postal questionnaires only.

    Secondly, we present proportions of cerebral palsy, severe/moderate sensory disabilities, cognitive deficiencies, behavioural difficulties, developmental coordination disorders, and level of neurodevelopmental disabilities by gestational groups. To better describe the group with mild disabilities and allow international comparisons, the different co-occurrences of impaired domains are reported. Survival, survival without severe/moderate neurodevelopmental disabilities, and survival without such disabilities among live births are then reported by weeks of gestational age at birth.

    Finally, school assistance, use of complex developmental interventions, and parental concerns are described.

    Percentages are given with exact 95% binomial confidence intervals, and means with standard deviations. We present results for complete cases and after multiple imputation to account for selective dropouts and missing information at follow-up. Variables in the imputation model included both those potentially predicting non-response and those predicting outcomes (web appendix 2). We included a large number of predictors to make the assumption, missing at random, more plausible and ensure reliable imputed results. Data were imputed by chained equations using the SAS multiple imputation procedure.34 Predictive mean matching was used for continuous variables with missing values, logistic regression for binary variables, and multinomial regression for categorical variables.

    We generated 50 independent imputed datasets with 30 iterations each. Estimates were pooled according to Rubin’s rule.35 We compared gestational age groups using either a test for linear trend, with gestational age group as a continuous variable, or using a χ2 test for categorical variables. All tests used generalised estimating equation models to account for the non-independence of outcomes related to multiple births. The reference sample was weighted to account for the sampling method and socioeconomic characteristics related to identified selection biases present at inclusion and arising from subsequent attrition, to provide a representative population sample (web appendix 1). Factors associated with neurodevelopmental disabilities were studied using multinomial generalised estimating equation logistic regression models, children without disabilities being the reference. Model covariates were gestational age, sex, single or multiple pregnancy status, small for gestational age, and parental socioeconomic status. All tests were two-sided; a P value less than 0.05 was considered statistically significant. Statistical analyses were performed using SAS version 9.4 and R 3.6.1 software.

    Patient and public involvement

    Patients were not involved in setting research questions or outcome measures, nor were they involved in design of the study. Parents of children in the EPIPAGE-2 cohort showed overwhelming support for the study through continuing attendance for follow-up. Contact with parents is maintained through newsletters and websites (; National parents’ associations assist with the dissemination of results.



    Participation in the study from birth to five years’ follow-up is shown in figure 1. At age 5½, 4441 children were alive, and 3083 (69.4%) were included in the study. Children lost to follow-up were more frequently female, born at 32-34 weeks, and to mothers with lower socioeconomic status (web appendix 3). These children were also more often diagnosed with cerebral palsy when born at 24-26 weeks, and with an Ages and Stages Questionnaire score below the threshold for children born at 27-31 weeks. Children with postal responses only were more often born to mothers of lower socioeconomic status, and had similar neonatal characteristics, but higher proportions of severe/moderate visual disabilities than children with data collected during a clinical examination (web appendix 4).

    Fig 1
    Fig 1

    Study population. *No survivors were born at 22 weeks and only one survivor was born at 23 weeks, who was lost to follow-up at 5½ years. ELFE= Étude Longitudinale Française depuis l’Enfance

    Neurodevelopmental outcomes among children born preterm

    Neurodevelopmental outcomes are presented in table 2. At 24-26, 27-31, and 32-34 weeks’ gestational age, cerebral palsy decreased from 12.4% to 5.9% and 2.4% and mean (standard deviation) full scale intelligence quotient increased from 89.6 (16.6) to 93.6 (15.7) and 97.3 (15.0). Behavioural difficulties were reported less often as gestational age increased, from 12.0% at 24-26 weeks to 8.2% at 32-34 weeks. Proportions of children with developmental coordination disorders were 18.8%, 8.5%, and 5.0% at 24-26, 27-31, and 32-34 weeks, respectively. Severe/moderate neurodevelopmental disabilities were reported in 27.8%, 18.7%, and 11.6% of children at 24-26, 27-31, and 32-34 weeks, respectively, whereas proportions of children with mild disabilities were 38.5%, 35.7%, and 33.8%. Including behavioural difficulties and developmental coordination disorders in the definition of mild neurodevelopmental disabilities increased proportions by 5% to 7% among the different gestational age groups (web appendix 5). Most of these children had a combination of cerebral palsy GMFCS-1, and mild sensory or cognitive impairments; behavioural difficulties or developmental coordination disorders were seen only in 18.0%, 16.2%, and 15.1% of children born at 24-26, 27-31, and 32-34 weeks, respectively. Outcomes by week of gestational age at birth are presented in figure 2 for live births according to the level of neurodevelopmental disabilities, and in figure 3 for cerebral palsy, severe/moderate cognitive impairment, and severe/moderate disabilities for survivors at age 5½.

    Table 2

    Neurodevelopmental outcome measures at 5½ years by gestational age group among survivors in the EPIPAGE-2 study. Values are percentages (95% confidence interval) unless otherwise stated. Observed and imputed data*

    View this table:
    Fig 2
    Fig 2

    Survival at 5½ years, survival without severe/moderate neurodevelopmental disabilities, and survival without any neurodevelopmental disabilities by week of gestational age at birth, among preterm born children in the EPIPAGE-2 study. Data corrected for study design and respondent selection. *Severe or moderate cerebral palsy (Gross Motor Function Classification System level 2-5), vision (bilateral binocular visual acuity <3.2/10), hearing (unilateral-bilateral hearing loss ≥40 dB not corrected or partially corrected with hearing aid), and full scale intelligence quotient less than two standard deviations below the mean of the reference sample born at term (web appendix 1). †Includes no cerebral palsy, no vision or hearing disabilities, full scale intelligence quotient greater than or equal to one standard deviation below the mean, no developmental coordination disorder, and no behavioural difficulties (table 1). CI=confidence interval

    Fig 3
    Fig 3

    Outcome at 5½ years by week of gestational age among survivors in the EPIPAGE-2 study. Data corrected for study design and respondent selection. *Full scale intelligence quotient less than minus two standard deviations, with cut-off point of the distribution related to the reference sample born at term (37-41 weeks; web appendix 1). †Severe or moderate cerebral palsy (Gross Motor Function Classification System level 2-5), vision (bilateral binocular visual acuity <3.2/10), hearing (unilateral-bilateral hearing loss ≥40 dB not corrected or partially corrected with hearing aid), and full scale intelligence quotient less than two standard deviations below the mean of the reference sample born at term (web appendix 1). CI=confidence interval

    Education and use of complex developmental interventions

    School enrolment was almost universal across gestational age groups, but proportions of children in a mainstream class without support increased with gestational age, from 72.7% to 86.0% and 93.3% at 24-26, 27-31, and 32-34 weeks, respectively (table 3). School support in mainstream class or special school increased with the level of neurodevelopmental disabilities but was not universal in children with severe/moderate disabilities and was also seen in children with mild, or without, disabilities. Considerable proportions of children had at least one complex developmental intervention during the year before the evaluation. Use of complex developmental interventions increased with the level of neurodevelopmental disabilities but was reported at 24-26 weeks in around half of the children with mild disabilities and one third of those with no disabilities. Among children with severe/moderate neurodevelopmental disabilities, 30% to 40% had no interventions. Finally, requests to the Maison Départementale des Personnes Handicapées were reported for 31.1%, 17.1%, and 7.8% of children born at 24-26, 27-31, and 32-34 weeks, respectively.

    Table 3

    Education and developmental interventions at 5½ years by gestational age group among survivors in the EPIPAGE-2 study. Values are percentages (95% confidence interval). Observed and imputed data*

    View this table:

    Parental concerns about development

    Parental concerns about development increased as gestational age decreased and with increased level of neurodevelopmental disabilities (table 4). Behavioural concerns were the most common in all gestational age groups. Around half of parents of children with mild disabilities, and a quarter of parents whose children had no disabilities, reported concerns in at least one domain.

    Table 4

    Parents’ concerns about child development at 5½ years by gestational age group among survivors in the EPIPAGE-2 study. Values are percentages (95% confidence interval). Observed and imputed data*

    View this table:

    ELFE reference sample

    After weighting, maternal level of education, maternal socioeconomic status, and paternal socioeconomic status were slightly higher in the reference sample than in the total ELFE cohort (web appendix 1). Parents of children in the ELFE reference sample reported the child’s health as being “good” less often than in the total ELFE cohort. They also reported more behavioural difficulties and greater use of complex developmental interventions, with psychological/psychiatric interventions and speech therapy being the most common.

    Factors associated with neurodevelopment disabilities

    Rates of neurodevelopmental disabilities were higher as gestational age decreased, in boys, in children born small for gestational age, and with decreasing parental socioeconomic status (table 5). In families with the lowest socioeconomic status, the adjusted odds ratio for children with mild and severe/moderate disabilities in comparison with children with no disabilities were, respectively, 3.16 (95% confidence interval 2.35 to 4.25) and 8.80 (5.97 to 12.96).

    Table 5

    Factors associated with neurodevelopmental disabilities at 5½ years in the EPIPAGE-2 study among the 4440 preterm children survivors born between 24 and 34 weeks’ gestation. Multinomial regression models with multiple imputation

    View this table:


    Outcomes at age 5½ in this contemporary French national cohort of children born preterm in 2011 indicate that preterm birth continues to pose a large burden for families, healthcare, and educational systems. Although the rates of severe/moderate neurodevelopmental disabilities decreased with increasing gestational age, around 35% of the moderately to extremely preterm born children had mild disabilities. These children often used special care or educational services. A considerable proportion of parents had concerns about their child’s development, particularly about behaviour. These concerns warrant attention.

    Strengths and limitations of the study

    Strengths of the EPIPAGE-2 study include the population based national cohort design with prospective enrolment of a large number of infants born not only extremely preterm but also very and moderately preterm. We used standardised developmental assessments to permit comparisons with other cohorts. Motor, sensory, and cognitive outcomes were used to describe rates of neurodevelopmental disabilities. Developmental coordination disorder and behavioural difficulties were included in the definition of mild disabilities to better reflect the complexity faced by preterm born children and their needs, which are rarely limited to intellectual deficits.36 We also report extra educational assistance and use of complex developmental interventions as they might have an important effect on family functioning37 and the need for health resources. Finally, scores for all tests are reported with reference to a sample of contemporary, term born children, examined with the same protocol.

    The main limitations are the selection biases inherent in long term follow-up studies. Loss to follow-up among preterm born children particularly affected families with lower socioeconomic status, younger mothers, and children born at 32-34 weeks. We used multiple imputation to account for these missing data, resulting in an increase in cognitive disabilities in all gestational age groups, compatible with the impact of socioeconomic status on development.38 The reference sample was weighted to account for non-representativeness of the total ELFE cohort for the French population. Residual differences in outcomes were noted between the reference sample and the total ELFE cohort, however, suggesting selective participation of worried parents. Importantly, interpretations of trends across preterm populations are not affected by these concerns. We did not perform any comparisons between the preterm groups and the reference sample owing to the different methods used to take into account missing data in preterm groups and the reference sample. Owing to the reference sample selection bias, however, differences between preterm and term born groups are probably underestimated. Finally, assessors were not blinded to the gestational ages of the children and this can be considered as another limitation.

    Comparison with other studies

    Reporting outcomes for children born preterm is challenging not only because completeness of reporting is difficult to achieve,39 but also because outcome measures have to be meaningful for both health professionals and policy makers and for children and parents.13 Two recent international Delphi surveys have defined themes to be reported in follow-up studies of preterm born children in high income settings.4041 In the first, general gross motor and cognitive abilities as well as sensory impairments were recommended for inclusion in a core outcome set.40 The second study identified education and the effect of minor impairments as being top research priorities.41 Our definition of mild neurodevelopmental disabilities is in line with these recommendations. In comparison with the commonly used definition, this refined definition increased the proportions of children with mild disabilities but, interestingly, to a similar extent (around 5% to 7%) across gestational age groups. This allowed us to describe the complexity and potential burden of mild disabilities for children and families. Compared with previous cohort studies,2 including the first EPIPAGE study,42 proportions of children with behavioural difficulties were low in EPIPAGE-2. Owing to the selection bias of the reference sample, rates of behavioural difficulties were probably underestimated in preterm born children. Interestingly, a discrepancy was seen between proportions of parents reporting concerns about their child’s behaviour and the proportions of children with behavioural difficulties as identified by the strengths and difficulties questionnaire. This is in accordance with a study exploring parents’ perspectives of children born before 29 weeks,12 in which a large majority of parents had a favourable opinion about the personality and happiness of their child, but more than half, whose children were classified as having no neurodevelopmental disabilities, had concerns about their child’s development, notably their behaviour.

    Using our definition, survival without any neurodevelopmental disabilities at age 5½ increased from 10% at 24 weeks to 50% at 34 weeks with an unexpected decrease at 34 weeks that might be due to sampling fluctuations. Overall, rates of disabilities and school assistance among children born at 32-34 weeks are in line with other reports of children born moderately preterm.434445 In many countries, including France, these children are excluded from follow-up programmes.2846 Primary care practitioners need to be aware of the difficulties families might face and be alert to parental concerns in order to avoid delayed referrals to specialised services for these children. Although school support depends on educational organisations within individual countries, consistency in school achievement across international cohorts has been described.47 In EPIPAGE-2, the high proportion of children requiring extra educational assistance emphasises that preterm birth already has an important effect on educational systems at this early age. Educational professionals in the UK have reported poor knowledge about the needs of preterm born children, and most feel ill equipped to provide school support.48 This lack of understanding is probably true in France and other countries, emphasising the importance of disseminating resources to improve teachers’ knowledge 49 and thus enhance appropriate support for these children.

    The proportions of children receiving complex developmental interventions were high, even among children with no, or mild, disabilities. These interventions indicate a high investment from parents, which might affect maternal economic status.50 Exploration of the nature of these interventions is important as multidisciplinary collaboration and coordination of care between professionals is far from excellent, and improvements have been advocated by parents’ associations to prevent families falling through gaps in care and to avoid needless duplication of treatment.37 Surprisingly, receipt of complex developmental interventions was not universal for children with severe/moderate neurodevelopmental disabilities, with a lower uptake of interventions in children with cognitive disabilities alone or with medical complexity,5152 suggesting increased susceptibility to inequities in healthcare in these populations. In line with this, requests to the Maison Départementale des Personnes Handicapées might be considered low in comparison with rates of disability. The increased risk of neurodevelopment disabilities in families of low socioeconomic status should be recognised by screening programmes.

    Unanswered questions and future research

    Identifying developmental profiles to understand the aetiological pathways leading to neurodevelopmental disabilities and, therefore strategies for prevention or intervention, is an important area for research.53 Limited educational and health resources could focus on children most at risk, with follow-up according to a child’s risk profile, and on improving information for parents to avoid unnecessary anxiety. Furthermore, understanding parental concerns about their child’s development offers improved possibilities for identifying appropriate interventions.


    In this large contemporary cohort of preterm born children, we report development at age 5½ in a broad range of areas, the need for educational assistance, complex developmental interventions, and parental concerns about their child’s development. This global perspective is important when advising parents, health personnel, and teachers, and also when designing follow-up and intervention programmes for children born preterm. Extra educational assistance and complex developmental resources were often used, even for children born moderately preterm or with no, or mild, neurodevelopmental disabilities. Difficulties faced by these groups of children and their families should not be underestimated.

    What is already known on this topic

    • Children born preterm experience difficulties in many developmental domains, with consequences for education, adaptive behaviour, and parental concerns

    • Neurodevelopmental disabilities at preschool age are usually described using a composite outcome including cerebral palsy, sensory and cognitive deficits

    • Rates of co-occurring problems, such as developmental coordination disorders or behavioural difficulties, in preterm children usually classified as having no, or mild, neurodevelopmental disabilities, are not readily available

    What this study adds

    • In France in 2011, children born extremely preterm and also those born very and moderately preterm remain at high risk of severe/moderate neurodevelopmental disabilities

    • Inclusion of behavioural difficulties and developmental coordination disorders in the definition of mild neurodevelopmental disabilities describes in more depth the complexity of difficulties faced by this group of children born preterm

    • Educational assistance and complex developmental interventions were used in all groups of children born preterm and with varying levels of disability, including those with mild or no apparent disabilities at age 5½


    We are grateful for the participation of all families of preterm infants in the EPIPAGE-2 cohort study and for the cooperation of all maternity and neonatal units in France. We thank parents’ associations (SOS Prema, Collectif Interassociatif Autour de la Naissance (CIANE), Jumeaux et plus, among others) for their overwhelming support and involvement in the dissemination of the results. We thank the EPIPAGE-2 study group for its substantial contribution to the conception, design, and acquisition of data. Details on the EPIPAGE-2 writing group and collaborators are in the web appendix list.


    • Contributors: VP, LM-M, and P-YA had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. VP, MK, ASM, ST, and SM conceptualised the study and wrote the manuscript. LM-M performed the statistical analysis. VB coordinated data collection and had responsibility for technical support. M-AC and P-YA obtained funding and supervised the study. All authors contributed to the analysis plan and interpretation of the results and reviewed and approved the final manuscript. All members of the writing group were involved in the regional organisation for data collection and reviewed or approved the final manuscript. All the listed authors met the ICMJE criteria for authorship. VP is the study guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

    • Funding: This project has been funded with support from the French Institute of Public Health Research/Institute of Public Health, and its partners. Partners include the French Health Ministry, National Institute of Health and Medical Research (INSERM), National Institute of Cancer, and National Solidarity Fund for Autonomy (CNSA); National Research Agency through the French EQUIPEX programme of investments in the future (reference ANR-11-EQPX-0038); PREMUP Foundation; Foundation of France (reference 11779); Foundation for Medical Research (SPF20160936356); and hospital clinical research programme Epinutri (DGOS13-040). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

    • Competing interests: All authors have completed the ICMJE uniform disclosure form at and declare: support from the French Institute of Public Health Research/Institute of Public Health and its partners for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous three years, no other relationships or activities that could appear to have influenced the submitted work.

    • Ethical approval: Assessments occurred only after families had received information and provided written informed consent. This study was approved by the National Data Protection Authority (CNIL DR-2016-290) and by appropriate ethics committees (Consultative Committee on the Treatment of Data on Personal Health for Research Purposes, reference No 16.263; Committee for the Protection of People Participating in Biomedical Research, reference 2016-A00333-48).

    • Data sharing: The EPIPAGE studies are subject to a data sharing policy that can be downloaded from

    • The lead author (the manuscript’s guarantor) affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.

    • Dissemination to participants and related patient and public communities: Results will be sent to families through newsletters and webinar. Media for dissemination will be available on request for the study group. A press conference will be organised for public communities. Funders receive annual reports.

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

    This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: