Introduction
Preterm infants born before 33 weeks of gestational age (GA), with a very low birth weight (VLBW, less than 1500 g at birth), or with a severe anomaly at birth are at high risk of developing a wide range of disorders during childhood, including motor, cognitive, sensory and behavioural problems.1–3 These infants also have higher rates of intellectual disabilities, behavioural, social and emotional problems and learning difficulties,4–7 which persist into adulthood.8–10 The development of such health problems considerably increases the extent and complexity of care required, leading to a significant economic burden on healthcare systems.11 Given that the rates of preterm birth and low birth weight have been steadily rising in European countries,12 these challenges are of growing public health concern.
Effective follow-up is essential to detect potential adverse outcomes early and to implement interventions aimed at minimising long-term complications.13–15 An evaluation at 2 years of age is critical for identifying severe disabilities and sensory deficits; however, it is insufficient to capture more subtle developmental difficulties that may arise later.13–15 Additional assessments between 2 and 8 years are recommended to better characterise these issues and to allow timely intervention. Despite the recognised need for follow-up, several gaps remain in our understanding of the specific relationships between perinatal risk factors, developmental outcomes and the quality and optimality of follow-up.16–19 In practice, the lack of large, longitudinal, population-based cohorts collecting comprehensive data makes it difficult to optimise routine follow-up services, improve cost-effectiveness,11 20 design targeted intervention programmes18 and develop innovative tools, such as predictive algorithms, to assist healthcare professionals.21
In France, where the rates of preterm birth and of low birth weight each reached about 7.0% and 7.1% in 2021.22 The medical follow-up for high-risk infants is coordinated by regional health agencies and provided by multidisciplinary perinatal networks involving a range of health professionals from both community and hospital settings, including medical specialists, general practitioners, paediatric nutritionists, paediatric physiotherapists, paediatric psychologists and paediatric psychomotor specialists. In response to this, the regional SEV-IDF programme was established in January 2016 in the Île-de-France (IDF) region, the most populous of the eighteen administrative regions of France and comprising the Paris metropolitan area. This paper aims to describe the methodology used to establish and manage this cohort, the characteristics of the follow-up, the baseline characteristics of included infants as well as the new opportunities in research that are emerging from the SEV-IDF cohort.
Cohort description
Study design and population
The SEV-IDF programme is a regional initiative designed to monitor high-risk infants born in the IDF region from birth to seven years of age, employing a prospective multicentric open population-based cohort design. Centred on the French capital Paris, it is located in the north-central part of France. IDF is densely populated though it covers only about 2% of metropolitan French territory; its estimated 2020 population of around 12 million was nearly one-fifth of the national total and is characterised by large socioeconomic disparities.23 Each year, more than 160 000 births occur in the IDF region.24 The region is made up of 8 administrative departments and has 59 maternities with a neonatal unit, of which 15 have neonatal intensive care units (figure 1).
Geographical locations of the Île-de-France region (IDF) in France (A) and of the maternities with a neonatal unit and referring physicians participating in the “Suivi des Enfants Vulnérables d’Ile-de-France” (SEV-IDF) programme (B) with a zoom on the Paris area (C). The 59 maternities with a neonatal unit in the region perform the recruitment. A network of 567 referring private or public physicians performs follow-up. The level of the maternity with a neonatal unit is based on the availability of neonatal care: Level 2A maternities have a neonatal unit only, level 2B maternities have a neonatal unit including intensive care unit with the staff and equipment to care for infants born at more than 30 weeks of gestation and level 3 maternities have a neonatal intensive care and resuscitation unit. The urban level of the municipalities is determined using the urban unit concept of the French National Institute for Statistics and Economic Studies. The three urban unit categories are classified as rural (less than 2000 inhabitants), suburban (2000–49 999 inhabitants) and urban (>50 000 inhabitants).
Eligible infants are those who are alive at discharge from the 59 maternities with a neonatal unit in the region and belong to one of the following mutually exclusive study groups:
-
Preterm (GA<33 weeks): infants born before 33 weeks of gestation.
-
VLBW: infant born with a GA between 33 and 37 weeks and weight below the third percentile of the French AUDIPOG curve,25 or those born at more than 33 weeks with a birth weight less than 1500 g.
-
Encephalopathy: infants diagnosed with neonatal encephalopathy.
-
Other: infants with a severe congenital anomaly or those who are twins with a birth weight less than 1500 g.
The programme initiated a pilot phase in selected geographical areas before officially starting in January 2016. The SEV-IDF programme is still ongoing, so the cohort of infants is an open cohort enrolling new infants each day.
Global objectives of the SEV-IDF programme were to (1) early detect developmental anomalies up to 7 years of age and help to understand their multifactorial mechanisms, (2) improve prevention with standardised epidemiological data, (3) evaluate and optimise the routine follow-up of high-risk infants in the region and (4) gather a multidisciplinary group of experts to coordinate and develop both epidemiological and operational research in the perinatal area.
Follow-up
The referring physician collects birth and discharge data at the time of hospital discharge. A network of 567 trained physicians monitors the children at 4 months (visit can be made before discharge from neonatology), 1 and 2 years of corrected age, and 3, 4, 5, 6 and 7 years of age. For preterm infants, corrected age is defined as chronological age minus the number of weeks ahead of the expected birth date.
Shortly before discharge, a list of referring physicians is provided to parents, who are free to choose the physician who will follow their child. These referring physicians are private paediatricians or paediatricians working in hospitals or maternal and child health centres who have agreed to participate in the SEV-IDF programme. The six perinatal networks (online supplemental figure S1) covering the IDF region are responsible for coordinating all the professionals involved in the follow-up of the included children, as well as their recruitment and training to use the standardised evaluations of the SEV-IDF cohort. These networks are also responsible for identifying contacts in each maternity and neonatal unit to oversee inclusion and data collection, scheduling follow-up appointments and monitoring the adequacy of the follow-up. To carry out these tasks, a dedicated team was specifically set up in each network.
Data collection
A large set of data is collected at birth, at hospital discharge and during the follow-up for all included children. Table 1 shows a summary of the data collected by the SEV-IDF programme.
Summary of the collected data in the “Suivi des Enfants Vulnérables d’Ile-de-France” (SEV-IDF) programme
Sociodemographic characteristics
Sociodemographic characteristics are assessed at discharge and updated at each follow-up visit using standardised questionnaires administered by the physicians. The demographics covered for the child include the date of birth, place of residence, family structure, housing conditions, healthcare coverage and household economic resources. The mother’s demographics include mother age, marital status, the education level and the socioprofessional category. The father’s socioprofessional category is also collected. The socioprofessional categories are based on the French National Institute for Statistics and Economic Studies (Institut National de la Statistique et des Etudes Economiques) classification,26 grouped as follows: (1) high-grade professionals (eg, executives, professionals); (2) intermediate occupations (eg, technicians, mid-level supervisors); (3) employees or farmers (eg, administrative or service personnel); (4) working parent with unreported occupation; (5) parent on parental leave and (6) unemployed parent.
Pregnancy and neonatal characteristics
Data concerning pregnancy, such as use of assisted reproductive technology, tobacco and alcohol consumption during pregnancy (declarative information), use of antenatal corticosteroids, chorioamnionitis, induced delivery, mode of delivery, are systematically collected in the SEV-IDF programme.
Routinely collected data during the neonatal period include sex, GA, anthropometric data (weight, height and head circumference) at birth and at discharge, multiple birth and information about the child’s conditions, care, management and treatment during the initial hospitalisation, such as surgery at birth, postnatal corticotherapy, administration of monoclonal antibodies against respiratory syncytial virus (RSV), bronchopulmonary dysplasia (oxygen-dependent infants at 36 weeks GA), stage II or higher retinopathy and neurological evaluation (low, moderate or high risk detected during clinical examination or in neuroimaging for the preterm infants with GA <28 weeks). Details of the neurological evaluation are provided in table 2.
Details of neurological classification
Follow-up characteristics
Several pieces of data concerning the follow-up are recorded, including the maternity hospital at birth and discharge if there was any transfer during hospitalisation, the geographical location and level of the maternity, The perinatal network responsible for the infant at discharge, the characteristics of the physician providing the follow-up (including their public or private status, medical specialty and place of practice), as well as all key follow-up dates (hospitalisation, discharge, follow-up visits and the end of the follow-up). The reasons for the end of the follow-up (Moving, death, unsatisfactory follow-up, other reasons) are also recorded when available.
The referring physician conducts several clinical evaluations of the child’s development during specific visits, focusing on anthropometric data (weight, height and head circumference) and detecting any respiratory, sensory, growth and neuromotor abnormalities. The physician also uses various standardised age-specific scores to evaluate the child’s social interactions, behaviours, attention capacity, sleep and diet quality, as well as toilet acquisition and overall quality of life. Examples of these scores are provided in the supplemental materials (online supplemental tables S1, S2) at the 2-year and 4-year visits. Questions concerning schooling and childcare characteristics are also collected.
At 24 months of age, parents are asked to complete the Age and Stages Questionnaires (ASQ).27 The ASQ is a questionnaire composed of age-specific questions divided into five areas of child capacities: communication abilities, gross motor skills, fine motor skills, problem-solving abilities and personal-social skills.
Data management and analysis
This multicentre open cohort relies on a web-based system with data entered into a central database. The system provides security with protected access and complies with French safety policy. Data entry and validation is performed as a continuous process. For each included infant, the referring physician in charge of the infant enters data directly online. Each perinatal network continuously controls data completeness and validity associated with their geographical area and notifies the local medical team in case of discrepancies or incomplete data. Routine training of personnel is regularly organised. Automatic data quality controls are performed periodically to control for missing data and value ranges.
Inclusion rate is checked each year by comparing the number of infants included in the cohort with the number of infants born the same year, living in the IDF region and meeting the inclusion criteria. The number of eligible infants is calculated based on data extracted from the French Information Systems Medicalisation Programme, an exhaustive national hospital discharge database maintained by the French Technical Agency of Information on Hospitalisation. Each year, between 80% and 85% of eligible preterm (GA <33 weeks) infants were included in the cohort.
CIs, means SD and frequency distributions are routinely calculated for all measures. In future descriptive analyses, robust univariate and multivariable models could be used to identify the various factors influencing the studied parameters such as neurodevelopmental outcomes, sociodemographic disparities or risk of loss to follow-up. Due to the longitudinal nature of the study, statistical methods may be used when needed to address methodological challenges, such as generalised linear models or Cox models to assess time-to-event outcomes such as hospital readmissions or other adverse events. Artificial intelligence methods, such as machine learning algorithms, may be used to develop predictive models for early detection of neurodevelopmental anomalies or to identify children at risk of being lost to follow-up based on clinical and socioeconomic variables.
Dissemination
This project involves a network of health professionals composed of the IDF Regional Health Agency, the 6 perinatal networks, the 59 maternities with a neonatal unit in the IDF region, health professionals involved in caring for the children, researchers from the National Institute of Health and Medical Research (INSERM) and the regional e-health public operator (GIP SESAN). The SEV-IDF programme is funded by the regional health agency. The GIP SESAN manages the programme’s information system, implements recommendations from the regional health agency and perinatal networks, and processes underlying data. A scientific committee has been established to validate all scientific and operational projects from the SEV-IDF programme. Results are shared with caregivers, funders of the cohort, medical societies, researchers and are submitted to international peer-reviewed journals and presented at international conferences.
Patient and public involvement
This study was conducted without involvement from patients or the public in its design, conduct, reporting or dissemination plans. There are no plans to involve patients or the public in the dissemination of the study results.
Findings to the date
As of 31 December 2023, 21 175 infants were included in the SEV-IDF programme and about 2600 infants are included each year. Tables 2–5 describe the characteristics of all infants, preterm (GA <33 weeks), VLBW, encephalopathy and other high-risk infants included. All these analyses are regularly updated to track changes in the cohort from the SEV-IDF programme.
Sociodemographic characteristics of high-risk infants at baseline (N=21,175), “Suivi des Enfants Vulnérables d’Ile-de-France” (SEV-IDF) programme, 1 January 2016–31 December 2023
Pregnancy characteristics of high-risk infants at baseline (N=21 175),”Suivi des Enfants Vulnérables d’Ile-de-France” (SEV-IDF) programme, 1 January 2016–31 December 2023
Neonatal characteristics of high-risk infants at baseline (N=21 175),”Suivi des Enfants Vulnérables d’Ile-de-France” (SEV-IDF) programme, 1 January 2016–31 December 2023
Sociodemographic characteristics
The baseline sociodemographic characteristics of the high-risk infants and their parents are presented in table 3. More than 1 in 10 infants (n=2678, 12.6%) lived in a family living in Paris, and nearly 3 out of 4 infants (n=13 315, 72.8%) lived in a highly urban area. Less than 1% (n=63, 0.3%) had a mother aged under 18, while over a third (n=7795, 37.3%) had a mother aged 35 or over. More than 1 in 10 infants (n=2401, 12.5%) lived in a single-parent family. Most (n=15 679, 83.8%) had healthcare coverage and 2808 (15.0%) had only coverage for low-income individuals. More than 80% (n=14 623) lived in individual housing and less than 20% (n=2872) lived in another housing (near relative/reception centre or emergency shelter). Most (n=9594, 51.5%) had a mother with a high level of education compared with 5.3% (n=979) whose mother had no education or only a primary level education. More than one in five infants (n=4260, 22.0%) had a mother declared as unemployed.
Pregnancy characteristics
The baseline pregnancy characteristics are presented in table 4. 13.5% (n=2,225) were conceived with the use of assisted reproductive technology. Tobacco and alcohol during pregnancy were reported in 1348 infants (7.7%) and 138 infants (0.8%), respectively. Three out of four mothers (n=14 035, 74.8%) received antenatal corticosteroid treatment, 1 in 10 infants (n=1754, 10.2%) experienced chorioamnionitis during pregnancy. Nearly one in two infants (n=8868, 48.7%) had a medically induced preterm birth, and a caesarean section was performed in most deliveries (n=13 073, 64.0%).
Neonatal characteristics
The baseline neonatal characteristics are presented in table 5. More than half (n=11,082, 52.3%) of the included high-risk infants were male. Including all inclusion groups, 17.8% (n=3775) had a GA of less than 28 weeks. More than a third (n=7775, 37.0%) had a birth weight between 1000 and 1500 g and one in five (n=4357, 20.7%) had an extremely low birth weight (less than 1000 g). One infant in five (n=4222, 20.1%) was small for GA (<10th percentile for Fenton z-score28). More than three in four (n=16 233, 77.2%) were appropriate for GA ((10th–90th percentile for Fenton z-score)) and less than 3% (n=559, 2.7%) were large for GA (>90th percentile for Fenton z-score). The mean (SD) weight, height and head circumference at birth were 1515.2 g (683.1 g), 39.7 cm (5.0 cm) and 28.2 cm (3.4 cm), respectively. Multiple births occurred in 5639 (26.6%) of the high-risk infants. The length of initial hospitalisation was greater than 70 days for 4657 (23.8) high-risk infants. During this time, 1541 (7.9%) infants had a hearing examination that required monitoring, 455 (13.2%) had stage II or higher retinopathy, but over 80% (n=17 735, 83.8%) had missing data. One in 10 infants (n=1683, 10.2%) received postnatal corticosteroids, and less than half (n=9425, 44.6%) had an indication for anti-RSV monoclonal antibodies. Nearly 9% (n=1520, 8.9%) underwent surgery at birth and 15.4% (n=3250) had bronchopulmonary dysplasia. Neurological evaluations were classified as moderate risk for less than 8% (n=1669, 7.9%) of the included infants and high risk for 3.0% (n=645). More than half (n=9185, 55.1%) were breastfed at discharge.
Follow-up characteristics
The baseline follow-up characteristics are presented in table 6. Most (n=11 865, 56.7%) of high-risk infants were included in a level 3 (highest risk) maternity, while 39.2% (n=7981) were born outside of their maternity of inclusion. Follow-up was performed by a majority of hospital-based physicians (n=11 907, 64.4%), a private physician for one infant in four (n=5128, 27.7%), and other physicians for 7.9% (n=1461). Nearly 40% (n=6716, 39.8%) infants lived at a distance inferior to 3 km from their referring physician, while 16.8% (n=2832) lived at a distance superior to 10 km.
Follow-up characteristics of high-risk infants at baseline (N=21 175), “Suivi des Enfants Vulnérables d’Ile-de-France” (SEV-IDF) programme, 1January 2016–31 December 2023
Future plans
Currently, the SEV-IDF programme is investigating three main research areas. The first area is focused on quantifying developmental difficulties in high-risk infants and identifying associated risk factors. The scientific committee is working on the construction of scores that summarise the children’s neurodevelopment at key ages to allow overall quantification and monitoring of these developmental difficulties in the IDF region. Ultimately, the aim is to use these neurodevelopmental scores to detect abnormalities early. The second area focuses on defining evidence-based retention strategies to reduce the high numbers of lost to follow-up traditionally observed in pediatric cohorts. This involves investigations to identify individuals most at risk of discontinuation. The final area is to use all these results to develop innovative tools to assist health professionals in these tasks.
Strengths and limitations of this study
One of the strengths of the SEV-IDF programme is its prospective and open design, which has enabled the collection of objective data on neurodevelopment and growth in a large number of children from birth to 7 years of age since 2016. As of 31 December 2023, the cohort has included over 21 000 infants, giving it the statistical power to study a wide range of hypotheses on perinatal health and care in various subpopulations. The cohort is population-based, covering nearly 20% of the French population (>12 million people) in a territory with diverse socioeconomic and geographical conditions, making it more representative of reality than a traditional research cohort does.
It is now recognised that a wide variety of conditions influence the outcomes and the healthcare of high-risk infants.16 17 The SEV-IDF database contains more than a hundred variables describing a large set of conditions such as prenatal complications, perinatal characteristics and morbidities, socioeconomic conditions and the organisation and efficiency of the healthcare system. This wealth of variables will make it possible to study the influence of a multitude of factors as well as to develop studies taking into account multidimensional factors and to better quantify their relative impact on the health trajectory of children.
One limitation of the programme is the use of non-validated scores to assess various aspects of children’s health and development. While there are numerous standardised tests available for evaluating the health outcomes of preterm children, such as the Wechsler Preschool and Primary Intelligence Scale to determine IQ29 or the Bayley Scale of Infant and Toddler Development-III,30 they can be time-consuming and resource-intensive and are therefore not used in the cohort. Instead, future validation studies will be conducted to confirm the reliability of the scores used in the SEV-IDF programme.
Another limitation is the high level of attrition, which is greater than 30% at the 2-year follow-up visit and 50% at the 5-year follow-up visit. These rates are higher than those reported in other studies.31 32 This may be due to the operational nature of the SEV-IDF programme compared with traditional research cohorts. Another explanation could be that the IDF is the region with the highest number of migratory movements. This first phase, presented here, will enable us to work on the issue of loss to follow-up, thanks to elements of assessment and practice. Moreover, some collected variables have a large proportion of missing values. This is most likely due to the operational nature of the SEV-IDF programme too and to the lack of time available to referring physicians to complete all follow-up forms for each child. In future analyses, to limit potential selection bias resulting from missing data, it will be essential to use appropriate methods such as multiple imputations for missing-at-random covariates.
Although a large number of variables are collected in the SEV-IDF programme, some information is lacking, particularly regarding the pregnancy period. In the future, the scientific committee plans to augment the cohort data by matching it with the French National Healthcare database Système National des Données de Santé.33 This will provide additional health-related information about infants, their mothers and associated healthcare throughout childhood.
The last limitation is the potential for non-exhaustive inclusion, where some subgroups within the target population may not have been adequately represented. A previous study in the IDF showed that the inclusion rate for preterm (GA <33 weeks) infants in the SEV-IDF programme was over 80%, but further analysis is needed to fully quantify the exhaustiveness of inclusions in the cohort and consider any potential biases that may result from this.
Acknowledgments
The authors would like to sincerely thank (1) all the health professionals who contributed to the SEV-IDF programme, (2) the members of the scientific committee and their coordinators, (3) the Agence Régionale de Santé d’Île-de-France for funding and coordinating the programme, (4) the six perinatal networks for their ongoing support and investment in field coordination, (5) the GIP SESAN team for their work on data acquisition, modelling, validation and support and above all (6) the SEV-IDF infants and families who have trusted us.
