Pregnancy Outcome Prediction Study: transgenerational and adults review (POPStar) (MR1489)

The University of Cambridge is the data controller who will process data for this project (funded by the Cambridge Biomedical Research Centre); Pregnancy Outcome Prediction Study: transgenerational and adults review (POPStar). The Office for National Statistics (ONS) will also be a data processor, acting as a trusted third party to enable linkage of HES data and POPS data to educational data provided by the Department for Education.

The original POPS (Pregnancy Outcome Prediction Study), was a prospective cohort study that was performed within the Rosie Hospital. POPS recruited a cohort of women in their first ongoing pregnancy between 2008-2012. Women were recruited when a viable pregnancy was confirmed at their dating ultrasound scan and 4,212 women were followed prospectively through their pregnancies until birth. Meticulous phenotyping was performed of maternal characteristics, fetal growth, biological samples, and delivery parameters. The POPS birth cohort has already been used to draw important new conclusions about fetal growth, prediction of pre-eclampsia, and gestational diabetes. POPS participants gave consent for all data and samples to be stored and used for future analysis without the provision of individual feedback of results. The final POPS pregnancy ended in February 2013.

The study will look at 4,212 pairs of mothers and children about whom very detailed information was collected during the Pregnancy Outcome Prediction Study (POPS). The POPStar study is a new longitudinal follow-up study of the ‘historical’ POPS cohort, involving linking detailed pregnancy data of participants to their later-life health and educational outcomes. The data provided by this application will be the current health status of POPS mothers and children. The data provided will thus enable the researcher to answer their key research questions – for example whether childhood attendance at neurodevelopmental clinics, neurodevelopmental delay, and educational attainment is linked to specific patterns of poor growth in the womb.

The over-arching aim of this application is to explore new ways of interpreting pregnancy data to find out how this relates to long-term health outcomes in women and their children. The participants of the POP study have been extensively phenotyped during early life, and thus following up their long-term health represents a unique opportunity to better understand the influence of early life on health and disease, and potentially to develop interventions to improve health.

The aim of the research is to understand how pregnancy data can be used to predict future health outcomes for mothers and children. University of Cambridge researchers know that growth and development during pregnancy has an important influence on health, but do not fully understand the relationship between pregnancy parameters and health outcomes in later life. It would be strongly in the public interest to develop better understanding of how pregnancy data can be used to predict and potentially prevent adverse health and neurodevelopmental outcomes in mothers and children.

(i) Studying outcomes in children

For the developing baby, the pregnancy environment is a key determinant of later health outcomes. It is increasingly accepted that the intrauterine environment has a lasting effect on health in later life. Understanding exactly what patterns of growth are linked to specific future adverse health outcomes for individual children would allow the possibility of early intervention for children at risk.

(ii) Studying outcomes in mothers

Over 80% of women in the UK experience a full-term pregnancy during their lifetime. Pregnancy constitutes a relatively short period of challenge to a woman’s normal health during early/mid-adult life. This ‘stress-test’ has the potential to unmask underlying disease propensity by revealing subtle impairments in functioning that are compensated for under normal circumstances. The ability to use pregnancy data to stratify later-life health risk for women would give opportunity to screen, monitor, and intervene early for high-risk groups.

The researcher aims to determine the current vital status of mother and child (using the MRIS list clean). The research aims to determine the current health and well-being status of the cohort by accessing their hospital episode statistics data. HES data is required to determine current diagnoses and health status in the cohort (for example diabetes in mothers, or developmental delay in children). Linking this data to their pregnancy data (collected during the POP study) and educational data (provided by the Department for Education via ONS) will enable the researchers to answer the key research questions about how to predict health problems from early life.

Section 251 approval has been sought because it has been demonstrated in other cohorts that positive response rates to re-contact after prolonged periods can be low (in the order of 30%). This is unlikely to reflect the actual rate of objection to data linkage, but more likely to do with participant inertia as regards opting-in, or inability to trace participants directly. Given the demographics of the Cambridge population (a relatively affluent group with high proportions of employment in sectors that involve frequent relocation such as healthcare and scientific research), there are likely to be high levels of movement amongst the POPS cohort since recruitment. As such, it is highly unlikely that an alternative strategy, such as participant recall would enable the researchers to achieve their research aims. The initial planned analyses in POPStar are powered based on data being available for at least 70-80% of participants. Having data available for <70% of the cohort would render at least one of the primary study questions (the link between patterns of growth and educational under- performance) unanswerable due to under-powering.

Processing will be carried out with the sole aim of performing scientific research in the public interest in accordance with General Data Protection Regulation Article 6(1)(e), General Data Protection Regulation Article 9 (2) (j). The research is in the public interest because the data will be used:

- To better understand the links between pregnancy data and later health outcomes in both mother and child

- To unravel the underlying mechanisms of these links

- To investigate how these data can be used in clinical practice to predict and prevent later adverse health outcomes in mothers and children. It is hoped that, if these mechanisms and links are successfully demonstrated in the cohort, and predictive model can be developed and incorporated into clinical practice.

The University of Cambridge are requesting pseudonymised (by study number only) individual level data, limited to data items from HES that are not considered identifiable.

The University of Cambridge will provide a list of POPS participants for a ‘list clean’ performed by NHS England, using their MIDAS system, in order to apply the National Data Opt-Out prior to implementing the study specific opt-out policy.

The researchers at the University of Cambridge will then contact alive participants at the mother's last known address with information and the option for them to opt-out of taking part in the POPStar study.

After those who do not wish to participate and ineligible participants are removed, the University of Cambridge will provide the following identifiers (obtained from the POPS study) and unique study number for each eligible participant to NHS England:

- Full name (Mother and baby)

- Date of birth (Mother and baby)

- NHS number (Mother and baby)

- plus unique, non-identifying, POPStar study ID (Mother and baby)

Linkage of the cohort to Department for Education data will be completed on name and date of birth only.

No data other than these identifiers will be provided by the POPStar study to NHS England.

These identifiers will be used by NHS England to match individual participant data to requested items from the pseudonymised HES data.

The project will involve pseudonymised data (identifiable only by study number) from the 3 different sources involved being brought together in the ONS secure research environment and linked by unique study number. The 3 sources from which data will flow into the ONS environment are (i) the POP study data of intrauterine growth and other metrics (provided by the university of Cambridge), (ii) educational data, including special educational needs and key stage scores (provided by the Department for Education), and (iii) health data (provided by NHS England). Linking both NHS England and DfE data to POPS data is essential to accurately understand the impact of intrauterine development on both physical health and neurodevelopmental outcomes.

The flow of data back from NHS England to the POPStar study will be via ONS as a trusted third party. The data flow will involve NHS England dropping the provided identifiers and returning the requested individual HES data items identified only by POPStar study ID. These data will then be linked in the ONS secure data environment to pregnancy data (for example ultrasound growth data) held by the POP study (University of Cambridge), and to National Pupil Database data held by the Department for Education. All data entering the ONS environment from the study sources (NHS England, University of Cambridge, and Department for Education) will be pseudononymised only by POPStar study ID. The linked HES/POPS/DfE data will be held and processed within ONS - identified only by study number (POPStar ID), which is pseudonymised hence mitigating the risk of any re-identification. Data will not be matched to any publicly available data.

The agreement covers 3 years of data in order that the researcher can build up a reliable composite snap-shot of current health and neurodevelopmental status. The geographical spread of the data is determined by the movement of POPS participants since the end of the study.

Data will be housed via ONS as described, and there will be no other flow of these data. The University of Cambridge and ONS are data processors. The data will only be accessed by members of the project team who are substantive employees of the University of Cambridge and ONS approved researchers.

For security and resource reasons the SRS is a Managed Service. Equiniti ICS (based in Belfast) maintains the system, on behalf of the ONS SRS. They do so through encrypted (TLS1.2) VPN tunnel and Remotely Access (RA) the SRS. All Equiniti ICS administrators are SC cleared and have no access to any data. ONS SRS Research Support “Admin” staff only have permissions to carry out such tasks as creating users, updating patches, testing and installing software applications, arranging disaster recovery, ITHC for the SRS environment, closing SRS sessions down, i.e. all the SRS environment Admin maintenance - essentially they are “power users”.

Equiniti ICS nor any their staff process the data. Therefore Equiniti ICS is not considered to be a Data Processor.

The ONS SRS environment is an isolated system. It has no connectivity to the internet other than using it as a bearer to pass TLS1.2 encrypted image packages for a virtual desktop infrastructure (VDI), hosted on an accredited cloud server hosted by UKCloud Ltd on the mainland UK. UKCloud Ltd merely host the environment, they have no access to data. Therefore CloudUK Ltd is not considered to be a Data Processor.

CLOUD SECURITY

NHS England security has provided assurance regarding the use of the Office for National Statistics' Secure Research Statistics service (ONS SRS), hosted by UKCloud Ltd in this application. The Office for National Statistics has submitted a selection of security documentation to support the use of cloud storage. NHS England Security have reviewed the documentation and provided relevant feedback, where necessary. NHS England are satisfied that the documentation demonstrates the level of security and governance in place.

The Office for National Statistics have supplied evidence to support:

• The use of the Data Risk Model to assess the Risk Profile Class.

• Risk Management of the use of the Cloud for this data, taking into consideration Confidentiality, Integrity and Availability.

• The use of Pseudonymisation.

• Board level involvement in the Risk Management Process evidenced through Minutes of these meetings.

• Understanding of the Shared Responsibility Model

The Office for National Statistics have a very good understanding of the security controls available to them to provide the appropriate controls to secure data in the Cloud.

Using the Cloud, benefits from the inherited controls that cannot practically be replicated locally such as Physical Controls, Resilience of Systems, Power Supplies, Communications and Geographically dispersed Data Centres within a region.

Elasticity in provisioning is also a consideration that benefits organisations in managing workloads. The Cloud provider, UKCloud, will use UK Data Centres only.

The Office for National Statistics' Secure Research Statistics service (ONS SRS), are hosted by UKCloud Ltd in this application. Equiniti ICS maintain this system. Both organisations provide hosting arrangements only and have no access to any data. Neither of these organisations are listed as data processors because of this reason and are listed as storage locations only.

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 ie: employees, agents and contractors of the Data Recipient who may have access to that data).

The University of Cambridge expect to publish the results of the POPStar analyses in peer-reviewed academic journals via open-access publication routes. The resulting academic papers will be targeted towards journals that are widely read by the obstetrics and paediatric research and clinical communities, including AJOG, JAMA Paediatrics.

Two scientific publications, including new findings relevant to children with poor growth in the womb and mothers with diabetes in pregnancy, have been submitted for peer-review at high-impact journals.

All outputs will be at aggregated level with small numbers suppressed in line with HES analysis guide.

The researchers also plan to disseminate the results through communication with other academics, including conference presentations, such as the Society for Maternal and Fetal Medicine and the Society for Reproductive Investigation along with presenting the work at meetings aimed at both academics in the same field and in different disciplines.

The University of Cambridge will also engage in public engagement activities to inform the wider public about findings and encourage use in health policy formulation.

The specific public engagement activities will depend in part on the actual study findings and their relevance to particular at-risk groups. However, the researchers will plan a number of key events:

- dissemination of the study findings at the annual Cambridge Science Festival. Each year, the Festival welcomes visitors to hundreds of events and receives extensive national and local media coverage. Over 170 event coordinators organise talks, interactive demonstrations, hands-on activities, film showings and debates with the assistance of around 1,000 staff and students from departments and organisations across the University and research institutions, charities and industry in the eastern region.

- Public talks highlighting the research findings.

The researchers have previously show-cased the findings of the previous research in a talk at the Hay Festival, which is attended annually by >250,000 people (talk in 2019 by Dr Catherine Aiken).

- Ongoing engagement with special interest groups to whom the findings of the research are likely to be relevant, for example via the Autism Research Centre, who co-ordinate studies involving families affected by autistic spectrum disorders and who provided valuable input into our funding proposal.

Updates are planned on the study website of study news/findings and reminder of opt-out possibilities. Additional updates may be considered at other times, particularly if important findings are to be made public in the national media. The researchers will provide annual study updates individually to any participant who requests this and provides contact details. The researchers will also aim to disseminate key findings using press releases, and the University/departmental/clinical school social media presence.

The University of Cambridge hope that the project will lead eventually to the development of predictive algorithms for later-life adverse health outcomes, that can then be prevented or mitigated by treatment strategies. For this long-term aim, the researchers will develop a collaboration with University of Cambridge Institute of Public Health.

There are no anticipated commercial outputs.

Over 80% of women in the UK experience a full-term pregnancy during their lifetime. Pregnancy constitutes a relatively short period of challenge to a woman’s normal physiology during early/mid-adult life. This stress-test has the potential to unmask underlying disease propensity by revealing subtle impairments in physiology and metabolism that are compensated for under normal circumstances.

A few examples of this are already well known and in use in clinical practice. For example, women who develop gestational diabetes during pregnancy have a 7-fold greater risk of going on to develop type 2 diabetes later in life than women who were not diabetic during pregnancy. It is therefore now recommended in the UK NICE guidelines that women who have had gestational diabetes have yearly follow-up so that if diabetes develops it is detected early and treated. However, much less is known about how other types of pregnancy data can predict long term health outcomes in women. The ability to use other pregnancy data (for example uterine artery blood flow) to stratify later-life health risk for women would give valuable opportunity to screen, monitor, and intervene early for high-risk groups.

An example could be using data collected in the POP study regarding the growth of the baby during pregnancy to predict which women will develop high blood pressure later in life. Coding for diagnosis of high blood pressure in the eligible POPS mothers will be obtained via provision of HES data in this application. High blood pressure diagnoses will be linked directly with detailed fetal growth records by POPStar, and the relationship explored and defined. If a significant relationship exists as predicted, then it should be possible to define a clinically useful risk-prediction model for later-life high blood pressure based on pregnancy data.

High blood pressure affects 26% of women in the UK, and is the third biggest risk factor for premature death and disease. At least 10% of women aged >35 in the UK are known to have hypertension in an unselected population, rising to more than 25% by the age of 45. The ability to determine an individual woman’s risk of developing high blood pressure early in her adult life using the growth of her baby as a predictive factor would allow for intervention such as improving lifestyle and early detection of hypertension. The growth of a baby depends on the development of a placenta, which in turn depends on the adaptability of the mother’ s cardiovascular system. Therefore, it is hypothesized that women whose cardiovascular systems do not adapt well to challenges will have both poor growth of their babies in the womb and also a higher risk of high blood pressure later in life. Because so much is known about the POPS pregnancies, there is a unique opportunity to understand how health in the womb influences later health outcomes.

Growth and development during pregnancy has an important influence on health of both mothers and children, but do not fully understand the relationship between pregnancy parameters and childhood health and neurodevelopmental outcomes. It would be strongly in the public interest to develop better understanding of how pregnancy data can be used to predict and potentially prevent both adverse health and educational outcomes in children. Knowledge of the links between pregnancy data and the risk of future adverse developmental outcomes for individual children would allow early intervention for children at risk.

A key example would involve children who are born with a low birth weight for their gestation (small-for-gestational age; SGA). Children born SGA are known to have poorer educational outcomes in mid-childhood compared to children born at normal weights. Knowing which patterns of SGA growth leave children at highest risk of learning difficulties or low educational performance could lead to the development of interventions to support their education and learning to prevent disparities in attainment. However SGA has multiple aetiologies and the causative pathway of the association with neurodevelopmental delay remains unclear. In particular, children born SGA can be divided into those whose are constitutionally small and those who have experienced poor growth in the womb.

The researchers hypothesise that, by comparing different growth patterns in the womb to mid-childhood educational attainment, they could define patterns of growth that put individual children at high risk of learning difficulties, and hence provide opportunity for early intervention (for example with learning support). In order to ensure that the links between growth in the womb and educational attainment are not confounded by other associated physical health issues in childhood, it is necessary to link data across 3 sources: pregnancy data, educational data and childhood health data.

Aside from fetal growth, the researchers also have detailed information on other pregnancy characteristics, such as maternal serum hormone levels, placental RNA, and metabolomics data. These are also potential important predictors of later health and development that will be used with both health and educational data to improve understanding of how we can predict adverse outcomes.

This research is now in the process of identifying groups of children who are at high risk of not meeting expected educational standards in mid-childhood and could therefore benefit from early educational support.

The anticipated benefits of the project, achieved by the project team, will therefore be:

- To better understand the links between pregnancy data and later health outcomes in both mother and child

- To unravel the underlying mechanisms of these links

- To investigate how these data can be used in clinical practice to predict and prevent later adverse health outcomes in mothers and children. It is hoped that, if these mechanisms and links are successfully demonstrated in the cohort, and predictive model can be developed and incorporated into clinical practice.

These benefits will be realised by the POPStar project team working at the University of Cambridge (the data controller who will process data). The benefits will be measured according to the published and presented academic research outputs. It is anticipated that initial outputs will be achievable within 2 years of obtaining project data.

The primary benefit to health and social care has been to generate new insight into the natural history of intrauterine growth restriction and the impact of early term delivery on later educational achievement. This research provides evidence around how babies with poor growth in the womb should be managed, at what stage they should be delivered, and what interventions should be offered to maximise their long-term health and educational outcomes. This work has resulted in new knowledge, published in peer reviewed journals, about the long-term outcomes of children whose mothers experienced common complications during their pregnancies, including anaemia, gestational diabetes, and high BMI.