Tracking the Impact of Gestational Age on Health, Educational and Economic outcomes: a Longitudinal Record Linkage Study (TIGAR)

The University of Oxford requires access to a pseudonymised extract of HES data which has previously been linked with birth registration data by City, University of London for a separate project. The University of Oxford required the linked data for part of the analyses required for the TIGAR study (Tracking the Impact of Gestational Age on Health, Educational and Economic outcomes: a Longitudinal Record Linkage Study) which is being conducted by a multidisciplinary team of researchers, led by the National Perinatal Epidemiology Unit (NPEU) at the University of Oxford.

The overall purpose of the TIGAR study is to investigate the effect of gestational age on health, educational and economic outcomes up to age 11 years. The intention was to link the data with education data. However, approval to perform that linkage was not granted under the previous iteration of this Data Sharing Agreement (V0.10). The University of Oxford are using an extract of the linked HES and birth registration data for the investigation of the effect of gestational age on health and economic outcomes only. The rest of this application refers to the health and economic analyses only.

Section 251 approval (ref: 15/CAG/0196) allows access to personal identifying information for the purposes of linkage and for checking linkage to produce a clean dataset.

The legal basis for the processing and storage of personal data for TIGAR is that it is ‘a task in the public interest (article 6e) and that sensitive personal data is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes (article 9 (2) (j)). The dataset released to the research team was ‘pseudonymised’, meaning that the team has no access to identifying data such as name, date of birth or NHS number.

The organisations involved in this study are ONS (data processor) and the University of Oxford. ONS hold the pseudonymised data within the secure SRS facilities. The researchers from the University of Oxford are performing their statistical analyses within the ONS SRS.

The linked data which is stored at ONS includes data from birth registrations, NN4B (National Number for Babies), HES (birth records and subsequent hospital admissions) and a variable that includes the age at death in months for those children who have died. Under another Data Sharing Agreement (DARS-NIC-10094-P6P4B) selected researchers from the TIGAR study team have had access to the linked data, including identifying details, at the secure ONS SRS facilities (previously known as 'VML') in order to check the matching between birth registration and HES data. This process is complete and a 'clean linked dataset' has been produced, and the University of Oxford have received a pseudonymised extract of data which is currently being analysed (i.e. statistical analyses for health and economic aspects of the study) at the secure ONS SRS facilities for the purposes of statistical analyses, the results of which would be disseminated at conferences and published in journal articles.

The reason for undertaking the work described is that it forms part of the TIGAR study. TIGAR is being funded by the UK Medical Research Council. It is being led by the NPEU, University of Oxford. TIGAR has had input from the support group 'BLISS for babies born too soon, too small, too sick' and the National Childbirth trust. Two Patient and Public Involvement Groups provide patients' and parents' perspectives on all aspects of the project (including the health and economic aspects described here) and will help guide the dissemination of the findings. An Advisory Group of experts (academic, clinical, support group) has also given input on the statistical and health economic analysis plan for the pseudonymised data under this Agreement.

The rationale for the TIGAR study is as follows. A typical full term pregnancy lasts about 40 weeks and babies born before 37 weeks of pregnancy are classified as preterm. Preterm birth (PTB) is a major cause of long term loss of human potential. Most preterm babies survive and do very well in the long term, but as a group they have an increased risk of health and neurodevelopmental problems in childhood and adulthood. Even babies who are born just a few weeks early may have more challenges in childhood than those who are born after a full term pregnancy.

Detailed information is needed on the typical health trajectories of children who have been born preterm. This will be used to inform clinical guidelines on which groups of preterm babies should be routinely followed up after birth. It will also be used to prepare health professionals for the type of difficulties preterm children may have as they are growing up and what support they may need. Information is also needed to facilitate the counselling of parents about the types of challenges that preterm children may have later in life - this will help parents be more prepared about what to expect in the future and when to seek help from professionals. Finally, information on the costs associated with prematurity will be used by organisations who plan or deliver health services - this will help ensure that preterm children who need help are identified and given support.

Studies of long term health outcomes following PTB need to be based on large numbers. There have been several large studies of the long term effects of PTB, but their findings may not be generalizable to the current UK population. There is a need for a large UK study to assess the health and economic outcomes, across the whole gestational age spectrum, in a population-based cohort of babies born in the 21st century.

Only data from 2004/5 to 2014/15 for individuals born in 2005 or 2006 will be processed.

The specific objectives of the TIGAR study (in relation to health and cost outcomes only) are:

1. To estimate the effect of gestational age, across the full spectrum, on hospital admissions in children up to age 11 years.

(This will be done by analysis of a pseudonymised extract of the linked data at the ONS SRS).

2. To describe the trajectory of these outcomes in different gestational age-groups of children using Zgrowth curve modelling[ methods. (This will be done by analysis of a pseudonymised extract of the linked data at the ONS SRS).

3. To determine whether these effects vary according to markers of socio-economic deprivation and whether there is a gestational age threshold beyond which the impact of gestational age is outweighed by the effects of socio-economic factors. (This will be done by analysis of a pseudonymised extract of the linked data at the ONS SRS).

4. To estimate the economic costs for hospital services in different gestational age-groups during the first 11 years of life, and the incremental costs associated with varying levels of prematurity. (This will be done by analysis of a pseudonymised extract of the linked data at the ONS SRS).

The sole data controller is the University of Oxford as they have sole control of how the data is held and processed. The Office for National Statistics is defined as a data processor as they process the linkage as defined by the study group at the University of Oxford.

The co-investigators at the University of Warwick, University of Leicester and City University do not have access to the pseudonymised TIGAR datasets which are being analysed at ONS , and therefore are not data processors. Their role in the project is to help develop the protocol and analysis plan for the study, to help interpret the results (based on the analysis of the TIGAR data at the ONS) and to help write the journal articles. Whilst providing valuable expertise the co-investigators have no ability to determine the purpose and means of processing the data.

Under a separate Data Sharing Agreement (ref: DARS-NIC-10094-P6P4B) birth registration data for all births in England and Wales in 2005-2007 have already been linked with the NN4B and HES maternity datasets by two of the TIGAR co-investigators. As part of an NIHR-funded project, they have updated this dataset to include births in 2008-2012 and linkage to subsequent HES data about admissions of the children to hospital for in-patient or day case care (data from 2004-05 to 2014-15). A subset of these datasets, which includes birth and hospital records for babies born in England from 1st January 2005 to 31st December 2006, is the 'master' dataset for the current application and is referred to as 'Baby Cohort'.

As part of the approval process for the current application, which was initially approved by IGARD in July 2018, it was required that the latest national opt-outs should be removed from the master dataset (‘Baby Cohort’) before a pseudonymised extract was created for statistical analysis. In order to apply the latest national opt-outs, it was necessary for the following steps to take place before access was granted:

1. ONS supplied a list of pseudonymised HESIDs for all individuals born in 2005 or 2006 from the data it currently held (under Agreement DARS- NIC-10094-P6P4B) to NHS Digital;

2. NHS Digital decrypted these IDs to reidentify the individuals and apply the latest national opt-outs;

3. NHS Digital re-encrypted the IDs using the same encryption (pseudonymisation) key as before and supplied to ONS a list of pseudo-HESIDs omitting the IDs of any individual who had registered a national opt-out.

4. ONS extracted the pseudonymised data linked with the returned pseudonymised HESIDs only and made that data available for the researchers from the University of Oxford to access within the SRS. This included all HES episodes from 2004/05 to 2014/15 for any individual born in 2005 or 2006 whose pseudonymised HESID was in the file returned by NHS Digital.

5. ONS created two unique pseudo-new identifiers (e.g. random numbers) for each child. These are known as tigarid and linkid. One of the pseudo-IDs (linkid) is intended for use by ONS to facilitate future data linkages and the other (tigarid) will facilitate future linkages with the extract being supplied under this Agreement. No additional data linkages are authorised under this Data Sharing Agreement and a separate application would be made to NHS Digital to undertake further linkage if/when required.

It may be theoretically possible, with multiple cross-tabulations, to try to identify an individual child or family that has a rare combination of characteristics, but the TIGAR team will have no requirement to do so and will never seek to do this

This process took place in the autumn of 2018 and the pseudonymised extract of the data was released for analysis within the ONS SRS facilities in January 2019. This extract is being analysed by the TIGAR team at the ONS SRS in order to achieve the specific objectives described above.

The data will only be accessed by individuals within the TIGAR study team for the purposes described, all of whom are substantive employees of the University of Oxford.

All organisations party to this Agreement must 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 - i.e. employees, agents and contractors of the Data Recipient who may have access to that data).

The specific outputs from the pseudonymised extract of the linked database will be scientific papers, conference presentations, and a lay summary/report:

The scientific papers and conference presentations will be aimed at the clinicians who provide obstetric, neonatal or paediatric care to those affected by birth before full-term. Results will also be disseminated to those involved in resource planning, policy and service provision and the development of interventions in terms of populations targeted and the content and timing of delivery. This includes services at a national (e.g. NHS) and local level.

The conferences will be chosen depending on what the key findings are and to which target audience. They will be most relevant in order to maximise the potential benefits of the research. Possible conferences include the European Congress of Perinatal Medicine (targeting obstetricians and neonatologists), the European Society for Paediatric Research (targeting paediatricians and those involved in child follow-up) and the Society for Social Medicine (targeting researchers in public health, epidemiology and health economics).

{ The lay summary will be disseminated to families affected by birth before full-term using Patient and Public Involvement (PPI) groups (parents, National Childbirth Trust (NCT) representatives) and other stakeholders such as ZBliss[ (a charity that supports those affected by preterm birth).

It was previously expected that the first paper would be drafted and submitted for publication in late 2018 / early 2019, with further papers published between 2018 and 2020, based on the assumption that access to the pseudonymised data would be granted in the summer of 2018. Access to the pseudonymised data was actually granted in January 2019. The analysis is still in progress and it is anticipated the analysis will continue until the end of July 2020.

The scientific papers will be published in Open Access journals and the report and lay summary will be freely available as pdf documents which can be downloaded from the TIGAR website (www.npeu.ox.ac.uk/tigar). When each paper is published, information will be posted on the NPEU website (https://www.npeu.ox.ac.uk/) and the TIGAR team will liaise with the University of Oxford Press Office to help disseminate to a wider audience via media interviews, etc. Finally, input will be sought from the TIGAR Advisory Group and PPI groups regarding other methods of dissemination, particularly targeting those who would be most interested in the findings.

All of the above outputs will contain only aggregate level data with small numbers suppressed in line with the HES analysis guide.

The beneficiaries of the research will be:

i) Children born before full-term and the parents and families who care for them.

ii) Those who provide obstetric, neonatal or paediatric care to those affected by birth before full-term.

iii) Other stakeholders who support families affected by birth before full-term. Stakeholders such as ZBliss[ (a charity that supports those affected by preterm birth) have already been identified and Bliss has agreed to support the project. The University of Oxford will use their networks to identify a wider group of stakeholders in health, who can provide input to the project (via the two advisory groups) and ensure that they benefit from the research.

i) Those involved in resource planning, policy and service provision (such as neonatal and paediatric healthcare services, public health services) and the development of interventions in terms of populations targeted and the content and timing of delivery (for example, whether it is important to focus on children born very preterm, or socially deprived preterm children). This includes services at a national (e.g. NHS) and local level.

For i)-iv) it is difficult to estimate what the exact benefits will be and when they will occur as this will depend on what the results show. However, possible examples might be that particular groups of pregnant women are advised to have induction of labour because the baby is not growing as well as it should be, or that particular groups of premature babies should be routinely followed-up and assessed by doctors and other health professionals. The pathway for putting in practice any findings or recommendations to change clinical practice would be the inclusion of the results in national clinical guidelines such as NICE and the Royal College of Obstetrics and Gynaecology Greentop Guidelines. The relevant charities such as those who support parents of premature babies may also add this new information to their website.

The benefits will start to accrue after the journal articles have been published and after we have disseminated the findings at conferences or via our patient support groups such as 'Bliss'. Publication and other dissemination will start to happen from late 2019 and will continue through 2020, possibly into 2021. We will measure impact by the number of times our work is mentioned by other researchers, whether it is included in any evidence statements or clinical guidelines, and whether our results are mentioned on other platforms (e.g. Bliss website and other websites used by parents of preterm children and the clinicians who look after them). The timeframe for these impacts might be several years after the results have been published.

A good start to the analysis of this large and complex data has been made, but the analysis is still in progress and the papers are not yet complete. The yielded benefits will accrue after publication and dissemination of the findings.