Educational outcomes in children born after assisted reproductive technology; a population-based linkage study

Infertility problems are common, with approximately 1 in 6 couples experiencing difficulty in conceiving children naturally. Fertility treatments can help many of these couples, and their use is rapidly increasing. Fertility treatments in which either eggs or embryos are handled are called assisted reproductive technologies (ART), and they include in vitro fertilisation (IVF) and other related techniques. Nearly 1 in 50 children in the UK are born to parents who have benefited from fertility treatments. There remains outstanding a major concern however of whether children born after assisted reproductive technologies (ART) are at higher risk of developing learning or behavioural problems.

There are biologically plausible reasons for increased vigilance regarding the development of children conceived after ART. These procedures involve the handling of eggs and embryos outside of the body at a vulnerable period of early human development, which could impact upon development of the nervous system and brain (neurodevelopment) in children conceived in this way. ART also carries increased risk of multiple births, premature delivery, and low birth weight, all of which are adversely associated with neurodevelopment.

The long-term cognitive and behavioural development of this increasing population of children has not been adequately investigated to date, because existing studies have been small, and have not included adequate comparison groups. A recent systematic review summarised that ‘there is insufficient evidence to conclude whether the long-term neurodevelopment of children born after ART is comparable to that of spontaneously conceived children’. The European Society of Human Reproduction and Embryology has emphasised that high-quality research is essential to understand whether or not this increasing population of children are at higher risk of developing any problems as they grow and develop, so that couples considering ART can receive appropriate and reliable information, and so that any problems in children can be identified and managed early.

University College London (UCL) are therefore conducting a population-based cohort study (a research study in which group(s) of individuals are followed over time) to compare educational outcomes in children born after ART procedures with two control (or comparison) groups of naturally conceived children: 1) naturally conceived siblings of the ART children, and 2) unrelated children chosen at random from the same schools (school-matched controls).

A parallel study conducted by UCL investigating children’s health following ART has identified a cohort of children born following non-donor ART procedures (ART procedures which did not involve the use of donor sperm or eggs) throughout England between 1992-2009 (n=86,064), as well the naturally conceived siblings of these children (n=23,299). These cohorts were identified by linking data from the Human Fertilisation and Embryology Authority (HFEA) database with the Office for National Statistics (ONS) birth records. This data is securely held by NHS Digital, with each individual’s records in the dataset pseudonymised using a unique member number (UMN). Pseudonymisation is a technique which helps to protect the personal information of data subjects by replacing information in a data set that identifies an individual with a reference number.

For this study, the proposal is to explore educational outcomes in this cohort of children born following ART procedures as well as their naturally conceived siblings. This will be achieved by linking the data for these children (held by NHS Digital) with the National Pupil Database (NPD), which is held by the Department for Education (DfE). The NPD contains detailed information about the educational achievement of all pupils in state sector schools and sixth-form colleges in England. Following change in legislation with the Digital Economy Act 2017, which aims to enable and facilitate the secure use of data from across the government sector for research, access to NPD data is now being made available for research purposes through the ONS Secure Research Service (ONS-SRS). The NPD will also be used to identify a second control (or comparison) group of unrelated (non-sibling) school-matched controls for the ART Cohort (unrelated children chosen at random from the same schools as the ART conceived children). The DfE will transfer the identifiers (including names, date of birth, sex, and postcode) of the unrelated school-matched controls to the ONS. The ONS will then identify information related to the birth of these children (birth weight, multiple birth status, and maternal age) from the national birth records. This birth information is important to ensure that the educational outcomes in these school-matched controls can be meaningfully compared with the ART conceived children. The ONS will then pseudonymise this birth information for the school-matched controls, and deposit it into the ONS-SRS where the pseudonymised data will be analysed by the UCL research team.

Through the linkage of several large existing national datasets, the proposed study will address an important gap in scientific knowledge regarding the educational achievement of children born following ART. It has a number of important methodological strengths compared to existing studies: much larger size, follow-up of children throughout the full age range of school education in England (4 to 18 years), national coverage, and the inclusion of two control (or comparison) groups of naturally conceived children (siblings of ART children, and unrelated school-matched controls) to provide a more robust and reliable assessment of educational achievement in the ART conceived children.

Data linkage using existing national datasets is the least intrusive, most efficient, and only feasible methodology for adequately investigating this research question with a sufficiently large and representative sample size. The existence of the HFEA database in the UK, with the mandatory recording of all treatment cycles of ART undertaken nationally as a legal requirement, offers an internationally unique opportunity to study the health and development of children conceived following ART. Linkage of the national cohort of children born after ART with the NPD is the only way in which the HFEA database can be utilised to investigate the educational outcomes in these children.

Within the remit of this data sharing agreement, UCL specifically aims:

i. To compare educational outcomes among children born following ART with children born following natural conception.

ii. To compare the frequency of special educational needs (SEN) and school exclusion among children born following ART with children born following natural conception.

iii. To compare outcomes for specific types of ART (e.g. intracytoplasmic sperm injection vs in-vitro fertilisation; fresh vs cryopreserved cycles) and specific causes of infertility.

It is necessary to use data for the whole available cohort of children conceived following ART (born between 1992-2009) for this study. This is because the large size of the study cohort is a key methodological strength of this study, and is necessary for the study to deliver reliable and accurate results. The size of the study cohort is critical to allow the study to have the statistical power to adequately address the research question, by providing sufficient numbers of children in the ART conceived group and the sibling control group at each Key Stage level of assessment in the national educational curriculum. The size of the study cohort is also essential to facilitate planned subgroup analysis, which will compare outcomes separately depending on the type of ART treatment used, and depending on the cause of infertility. Under the terms of the Human Fertilisation and Embryology Act 2008, HFEA data on ART cycles carried out before 01/10/2009 can be used for research, subject to ethical approval, without explicit patient consent. Patients can withdraw consent, but only around 350 have done so. This makes the data prior to 01/10/2009 virtually complete. After this date, an opt-in system for research consent applies and the data is much less complete. The data after 01/10/2009 is not covered by Confidentiality Advisory Group s251 approval and is therefore not available for use in this study.

The principles of data minimisation will be respected by limiting the data transfer to the minimum individuals necessary (individuals within the study cohorts), by limiting the variables (or data items) transferred to the minimum necessary to facilitate the linkage and subsequent analysis (see ‘Processing Activities’), and by pseudonymisation of the data as soon as data linkage is complete (prior to data analysis by UCL). It is not possible to further minimise the data, as the size and geographic spread of the study cohort across England are essential for the study to have sufficient statistical power to adequately investigate the research question. The ART cohort includes all children born in England following IVF procedures between 1992-2009.

With respect to the processing of information flowing from NHS Digital, the Data Controller for the study will be University College London (UCL). The study purpose and design has been determined by the research team at UCL.

UCL, the DfE and the ONS will be Data Processors. The DfE are processing the data for their respective linkage step (combining data for the study groups with the National Pupil Database). UCL research staff will process the record level pseudonymised data after data linkage processes have been completed. The linked pseudonymised data will be held within the ONS-SRS for analysis by UCL research staff. The ONS-SRS is a system hosted by the ONS which will store the linked record level NHS Digital data. ONS staff check all output to be transferred out of the SRS to ensure that it is aggregated with small numbers suppressed (so that information regarding individuals is not disclosed), and to ensure that it cannot be combined with other data sources to identify individuals.

The main ethical consideration regarding the proposed data dissemination is the use of individuals’ data for research without individual consent being sought. Favourable independent review of the study proposal by a Research Ethics Committee (REC) and the Health Research Authority Confidentiality Advisory Group (CAG) has confirmed that use of individual data without consent is justified on the basis of the public interest of this high-quality research and the safeguards in place to maintain data confidentiality (pseudonymisation of the data before it is made available to the UCL research team for analysis, and the data security assurances for the data controllers and processors).

Access to personal identifiable data from NHS Digital will be restricted to a limited number of staff at the DfE who are experienced in handling sensitive data and who are bound by confidentiality agreements as part of their employment contracts and codes of practice. Robust security measures will be in place for the transfer and storage of personal identifiable and sensitive data, and the UCL research team will not have access to any personal identifiable data at any stage. The minimum necessary patient identifiers will be transferred between organisations for the purposes of linkage, alongside birth weight, multiple birth indicator and maternal age (with no transfer of other fertility, treatment or educational data between the organisations performing linkage).

The study has been funded by a research grant from the Nuffield Foundation. The Nuffield Foundation has no role in the design or conduct of the study, and will not be processing or accessing any data. One of co-investigators in the study team is based at the University of Oxford and Birkbeck University of London. They have an advisory role only, and will advise upon the statistical analysis and interpretation of the data. They will not have access to individual level data in the ONS SRS. The University of Oxford and Birkbeck University of London are not considered to be Data Controllers or Processors for the study.

The legal basis for the data processing proposed for this study is that of Public Task, as set out in Article 6(1)(e) of the GDPR. Universities are classed as public authorities for the purposes of data protection law. When carrying out research, as is proposed in this agreement, UCL will be carrying out tasks in the public interest in its capacity as a public authority. The legal basis for processing special category data is under Article 9(2)(j) of the GDPR (processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes). Appropriate safeguards will be in place when processing data in accordance with Article 89(1) of the GDPR. All data made available to the UCL research team will be pseudonymised using a unique member number (UMN). The UCL research team will not have access to individual identifiable data at any stage. Individual identifiable data will be accessible only to a limited number of DfE staff responsible for conducting the data linkage, within secure DfE internal systems.

Legal provision for the processing of identifiable data from the HFEA Register for the purposes of medical research is provided under section 33D of the Human Fertilisation and Embryology Act 2008. UCL have received approval from the HFEA’s Register Research Panel for the use of HFEA Register data for this project.

The study will involve the following data processing and linkage steps:

1. Linkage of the ART Cohort and Sibling Cohort to the NPD:

Individual level data held by NHS Digital under DARS-NIC-180665-GJMW5 regarding the ART and Sibling Cohorts will be securely electronically transferred to the DfE, for linkage to the NPD. Data to be transferred will include individual identifiers required to facilitate linkage (forename, surname, gender, date of birth, postcodes at multiple time points, and unique member number (UMN)), as well as existing variables in a separate pseudonymised bridge file that will be used as covariates in subsequent analysis (birth weight, multiple birth indicator, mother's month and year of birth). The DfE will not attempt to match individuals' birth information to their identifiable data. In order to optimise linkage to the NPD - which records postcode data yearly over the course of individuals’ period of school education - NHS Digital will transfer all available postcodes for individuals in the study cohorts from 2004 onwards (the period for which the NHS demographics dataset is available). Prior to data transfer, NHS Digital will exclude individuals who have submitted a National Data Opt Out.

Substantively employed staff at the DfE will perform the linkage to the NPD on-site in DfE internal systems, with development of the linkage algorithm in collaboration with the UCL research team. The DfE will link the ART children, and their naturally conceived siblings, to their educational outcome data (from the Early Years Foundation Stage Profile through to Key Stage 5), as well as to information regarding secondary outcome measures (special educational needs and school exclusion).

2. Identification of school-matched controls:

Staff at the DfE will identify unrelated (non-sibling) matched controls for the cohort of ART children, using a 1:1 matching algorithm developed alongside UCL. For each level of Key Stage outcome data available for each child in the ART Cohort, one matched control will be identified and randomly selected from the NPD database, with matching by school, age, and gender. A different matched control will be identified for each level of Key Stage outcome data available for each ART child. This is because children can change school during follow-up, and because a single control child may not have outcome data available at all of the Key Stage levels for which the matched ART child has data available. Linking the matched controls back to the ART Cohort file will facilitate a check that the matched controls were not conceived by ART. If any matched controls are found to be ART conceived, they will be replaced with an alternative randomly selected matched control who is not ART conceived.

UCL will request the following outcome and demographic data held in the NPD for children in the study cohorts; test scores for national ‘Key Stage’ pupil assessments at fixed points during school education between 5-18 years of age (Early Years Foundation Stage Profile through to Key Stage 5), special educational needs (SEN) status, school exclusion status, ethnicity, eligibility for free school meals, main language, and Income Deprivation Affecting Children Indices (IDACI) score.

3. Linkage to ONS birth records for school-matched controls:

In order to ascertain information regarding key potential confounding variables (birth weight, multiple birth status, and maternal age) for the unrelated school-matched controls, a minimum number of individual identifiers (names, dates of birth, gender, and postcode) for the unrelated school-matched control group will be securely transferred from the DfE to ONS for linkage to ONS birth records. The data linkage will be performed by ONS staff, within ONS internal systems. N.B. This data linkage step is outside of the scope of the data sharing agreement with NHS Digital, occurring under a separate agreement between UCL, the DfE and ONS.

4. Creation of final dataset for analysis:

Once data linkage processes have been completed, the data will be pseudonymised by DfE and ONS staff using the UMN. The pseudonymised dataset will contain demographic data (gender, age, age within academic year, ethnicity, area-based index of deprivation (IDACI), and eligibility for free school meals) as well as educational outcome data for children in all three groups (ART Cohort, Sibling Cohort, and unrelated school-matched controls). The pseudonymised datasets will be transferred from DfE and ONS internal systems to the ONS-SRS system to allow access by the UCL research team for data analysis. The bridge file of birth covariates (birth weight, multiple birth status, and maternal age) transferred from NHS Digital to the DfE, keyed on UMN, will also be onwardly transferred into the ONS-SRS and not retained by the DfE. The SRS is the ONS facility for providing secure access to sensitive detailed data for accredited Approved Researchers, and conforms to the NHS Data Security and Protection Toolkit. Named Approved Researchers are able to access data related to their project either through a secure online connection to the SRS system from a University computer, or at a physical desk in one of the ONS offices (designated Safe Settings).

Data regarding fertility treatment for the ART cohort is held by UCL in pseudonymised form (using the UMN), under a data sharing agreement with the HFEA. This will be securely transferred from UCL to the ONS-SRS. This data will include the type of ART used and the cause of infertility. The UCL research team will link this data to the demographic and outcome data for the cohort within the ONS-SRS system, using the UMN.

The UCL research team will conduct analysis of the final pseudonymised dataset through the ONS-SRS. At no time will UCL have access to identifiable data. Access to the data will be restricted to a limited number of named UCL research staff working on the project (both substantive employees of UCL and honorary contract holders). Named UCL research staff responsible for conducting the analysis for the project will complete the ONS Researcher Accreditation process, which involves specific training in the safe use of research data environments. They will sign and adhere to the ONS Accredited Researcher Declaration, and will be required to adhere to ONS data protection policies and procedures. This training in data protection and confidentiality through the ONS Researcher Accreditation processes, and the contractual agreements with the ONS to protect the data through the ONS Accredited Researcher Declaration, will apply to both substantive UCL employees and honorary contract holders working on the project.

In the data analysis, performance at each Key Stage level will be compared between the ART group and each of the two control groups (naturally conceived siblings and unrelated matched controls) using linear regression models. Because the data are matched, mixed effects linear regression models will be used. Regression models will also be used to analyze the secondary outcomes (SEN and school exclusion). Because these outcome variables are binary (yes/no) variables, logistic regression models will be used. Analyses will control for birth weight, multiple birth status, birth order, and age within academic year (by month). Analysis involving matched, unrelated controls will additionally control for ethnicity, maternal age at time of birth, postcode-linked social deprivation score (Income Deprivation Affecting Children Indices), and eligibility for free school meals (as a marker of poverty). Subgroup analyses will be performed for children conceived using different types of ART treatment (intracytoplasmic sperm injection vs in-vitro fertilization, fresh vs. cryopreserved cycles) and also for children whose parents have specific causes of infertility (female factor, male factor, both male and female factors, unexplained).

Individual level data will at all times be held within ONS systems, and at no stage will individual level data be transferred to or processed within UCL systems. All data to be transferred out of the SRS (the results of the analyses) will be checked by ONS staff to ensure that no individual level data, or potentially identifiable data, is transferred. Only aggregate level data, with small numbers (<10) suppressed, will be transferred out of the SRS system for publication. When the UCL research team access the pseudonymised dataset in the ONS-SRS they will be performing all analyses within the ONS-SRS system, and will only be extracting aggregated data.

Identifiers that the research team at UCL will have access to (ethnicity, gender, and month & year of birth) are important potential confounders to control for the analysis. The identification of research participants will not be possible from these variables. UCL does not hold any other information which could allow identification of individuals through interaction with the pseudonymised data that UCL will receive during the study (from NHS Digital, the DfE, and the ONS). There will be no requirement or attempt to reidentify individuals.

5. Data Retention:

Individual identifiable data for the study cohort will held by the DfE for a period of three months following completion of the data linkage. This will allow for DfE staff to investigate any data anomalies or discrepancies, should they be identified by the UCL research team during data cleaning and analysis. The DfE will securely delete all identifiable and birth covariate data for the study cohort after this three-month period. Only the pseudonymised analysis dataset will be retained after this point, to facilitate analysis by the UCL research team.

NHS Digital reminds all organisations party to this agreement of the need to comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract ie: employees, agents and contractors of the Data Recipient who may have access to that data).

Outputs are expected to include:

a) Peer reviewed publications in medical journals. The main results of the study are intended to be submitted for publication in a high impact general medical journal (such as The Lancet, The Journal of the American Medical Association (JAMA), or the British Medical Journal (BMJ)). Publications will be Open Access as per UCL policy, and freely available via both journal websites and UCL webpages.

b) National and international conference presentations. Proposed conferences for presentation include the European Society of Human Reproduction and Embryology (ESHRE) Annual Meeting (2023, date not available yet), and the American Society for Reproductive Medicine (ASRM) Scientific Congress (October 2023).

c) Report for the study funder (Nuffield Foundation), which will be publicly available via the study webpage on the Nuffield Foundation website.

d) Brief lay summary report, which will be made publicly available via the websites of stakeholder organisations (HFEA, Fertility Network UK, Nuffield Foundation, and UCL)

Outputs will contain only aggregate level data with small numbers (<10) suppressed in line with ONS and DfE policy.

Dissemination of the research findings to researchers and scientists will involve presentation at national and international conferences and publication in peer review medical journals, as detailed above.

Dissemination of the research findings to the public (key stakeholders being parents who have used or are considering ART) will be facilitated through existing collaborations with the HFEA and Fertility Network UK (the leading patient organisation supporting people suffering from infertility). The research project has already been informed by Patient and Public Involvement work, with UCL conducting a survey of parents of children born following ART, which determined that children’s educational potential was one of the most common concerns they had.

Dissemination of the research findings to a lay audience will be in the form of a brief research report and a video summary. Communication of the research findings to the public will be via the websites of stakeholder organisations (HFEA, Fertility Network UK, Nuffield Foundation, and UCL), and via the newsletters and social media channels (e.g. Twitter) of these organisations.

Research regarding fertility treatments, including previous work of the UCL research team, has attracted a high level of media interest, and the team anticipate that this will be the case for the proposed study. The team are acutely aware of the potential harmful effect of inaccurate or sensational reporting of research findings in this sensitive area, and the confusion and anxiety this can cause for couples and parents. The team will work closely with the HFEA, Fertility Network UK & UCL to co-ordinate press releases and ensure that information is conveyed accurately and responsibly.

The research team will commence data analysis as soon as the linked data has been made available. The team would anticipate that the process of data analysis, interpretation and report writing would take approximately 12 months. The team anticipate that the analysis will be completed and outputs generated in late 2023, although this estimate is dependent upon the timeframe for data access approvals being obtained and the data linkage being completed.

The HFEA regulates fertility treatment in the UK, with HFEA policy having the potential to influence the large number of fertility treatments performed each year - approximately 54,000 women received ART in the UK in 2018. UCL will provide the HFEA with a report of the study findings via the HFEA Register Research Panel, and anticipate that the results of the study may influence policy decisions by the HFEA. For example, there are concerns at present that some more invasive and expensive forms of ART are being over-used in some treatment centres, with wide variation in practice between centres. Should the results of this study demonstrate potential risks associated with particular types of ART, this may influence HFEA guidance regarding the use of different types of treatment, potentially saving costs for the health service and/ or distress for the recipients of ART. Any change in HFEA guidance is likely to directly influence the clinical practice undertaken by fertility clinics.

The HFEA have also articulated concern about the use of ‘add-on’ treatments by private fertility clinics, and the linkage methodology established by this study is expected to facilitate replication of the study in the future for evaluation of the safety of emerging technologies. Measurement of these potential benefits are intended to be through changes in HFEA guidance and policy regarding fertility treatment, and ongoing analysis of national ART treatment practices in the HFEA annual fertility treatment statistics.

Publication of the research findings in peer reviewed journals and presentation at scientific conferences is anticipated to enable dissemination of the results to a wider medical, educational, and scientific audience. This is intended to enable the results to influence clinical practice in fertility medicine, including beyond the remit of the HFEA such as internationally. It is also expected to help inform education providers as to whether this group of children may benefit from additional educational support, and enable the results to influence future academic research in this field.

UCL’s engagement with the HFEA and the media is anticipated to help disseminate the results of the study to the general public. The study’s findings are expected to be of considerable interest to families who have used or are considering ART. Previous research has shown that many parents who have used ART are concerned whether their children are at higher risk of developing learning or behavioural problems as a result of the fertility treatment. There are biologically plausible reasons for increased vigilance regarding the neurodevelopment of children conceived after ART, because of the potential for ART procedures to influence the nervous system during a vulnerable period of development. This important aspect of the health and development of ART conceived children has not been adequately investigated. As educational performance is a key measure of children’s neurodevelopmental progress once they reach school age, this study is expected to provide families with robust information that addresses these concerns.

Although UCL do not anticipate that the study’s findings would directly influence couples’ decisions about whether or not to use ART, couples using ART treatments experience considerable anxiety and psychological stress both during and after treatment. Should the study show there to be no difference in the educational attainment of ART conceived children compared to naturally conceived children, it is anticipated that this will provide considerable reassurance for families. The resulting reduction in anxiety experienced by parents of ART children has the potential to improve their wellbeing and family functioning. The study has the potential to benefit tens of thousands of families annually across the UK. Measurement of this potential benefit is likely to be through ongoing qualitative research and surveys exploring the experiences of people using ART, such as the HFEA national fertility patient survey.

Should the study show evidence of reduced educational attainment in ART conceived children, this will identify that these children may benefit from additional educational support. This could influence the decisions made by parents and education providers regarding the provision of such extra educational support for their children. It might also stimulate further research to explore the specific domains in which children may experience difficulties and benefit from support. Targeted extra support for children could have the potential benefit of minimising or overcoming any identified attainment gaps, thus improving the educational outcomes of these children. This has the potential to benefit a large number of children annually; over 18,000 children were born following ART in the UK in 2018.

Dissemination of the research findings to families using ART and to education providers is planned to be achieved via publicity through various stakeholder organisations, including Fertility Network UK, the HFEA and the Nuffield Foundation. Measurement of the potential benefits of this dissemination is proposed to be through qualitative research and surveys exploring the experiences of families using ART, as well as through follow-up studies of educational outcomes in children conceived by ART (for example by repetition of the linkage methodology used for this study).

The Confidentiality Advisory Group specifically considered the public interest of the research, and concluded: “The Group discussed the application and agreed that this defined a clear medical purpose which was in the public interest as high-quality research which was essential to understand whether or not this increasing population of children born following assisted reproduction therapy (ART) were at higher risk of developing learning or behavioural problems as they grow. This information would also enable individuals considering ART to receive appropriate and reliable information, and ensure that any problems in children can be identified and managed early.”

This is a new Data Sharing Agreement. No data has been disseminated by NHS Digital for this research study. There are therefore no yielded benefits to date.