Long term health outcomes for women and their children after assisted reproductive technologies: a data linkage study for England

Infertility affects 1 in 7 couples, and the number of babies using Assisted Reproductive Technologies (ART) such as IVF is increasing. Since 1991 over 200,000 children have been born in the UK as a result of ART, which now accounts for about 1-2% of births each year. Most IVF births result in healthy children, but there is good evidence that IVF pregnancies are at higher risk of poor perinatal outcomes (prematurity, low birth weight, a higher incidence of caesarean section), congenital malformations and some rare epigenetic syndromes. Cognitive, developmental and behavioural problems have also been identified, though the evidence is limited and findings are less convincing - in part, because studies in this area are difficult to conduct and sometimes flawed. Higher rates of other adverse outcomes, such as asthma and autism, have also been reported but, again, the evidence is sparse. Studies of health and developmental outcomes of children born after ART tend to focus only on early childhood, so that the long term health and development of ART children is still relatively under-researched.

There are also well-recognised risks to the woman at the time of ART treatment, and the negative impact of unsuccessful treatment on women and their partners has been documented. However, when treatment is successful the longer-term implications for the mother are often overlooked. The difficulty of becoming pregnant, perceived high-risk pregnancies and anxiety about birth can all have a lasting impact. The incidence of postnatal depression is higher after ART, and sufferers appear to be more vulnerable to later depressive episodes. Fears about increased cancer risks in women who have undergone treatment appear unfounded, but there remain unanswered questions regarding long-term impacts of ART on maternal physical and mental health, and use of health services.

This research seeks to address these issues, and provide evidence for the health and wellbeing of ART mothers and their children in the longer term. The project uses data that is collected by infertility clinics (on behalf of the Human Fertilisation and Embryology Authority) on all ART treatments in the UK since 1991, and links it to GP medical records for the women and their children held in the Clinical Practice Research Datalink mother-baby dataset. It is anticipated that over 280,000 mother-baby pairs will be included, of whom about 4,000 will have been born after ART. The oldest children will have been followed up for over 20 years. We will be exploring the health of children and their mothers using information that is recorded by the doctor or nurse practitioner when they visit their doctor's surgery, such as diagnoses of medical conditions, prescriptions for medication and referrals for specialist treatment. The increased use of health services, if any, among ART mother-baby pairs will also be assessed.

Currently the HFEA seeks consent from all couples undergoing ART to allow scientists to use their data in research, but less than half agree. Another aspect of this project will be to assess the effect of this low participation rate on the results of research using the HFEA data, and to explore ways that the effects of missing data can be reduced or removed. These projects all use routine data, but we will also be conducting a study that recruits couples undergoing treatment and explores why they choose to give, or to refuse, consent for research. The aim of this part of the study is to provide information to try and help increase consent rates in the future.

The research will be led by Claire Carson, an epidemiologist at the National Perinatal Epidemiology Unit, University of Oxford, who will be working with specialists in perinatal and maternal epidemiology (Jennifer Kurinczuk, Maria Quigley, Alistair Sutcliffe), qualitative research methods (Lisa Hinton), health economics (Oliver Riviera-Arias) and the management of missing data (Mike Kenward).

The main aim of this project is to assess the long-term health outcomes of mothers and children after the use of assisted reproductive technologies (ART) in England.

A linked dataset combining information from the Clinical Practice Research Datalink and the Human Fertilisation and Embryology Authority (HFEA) register will be created. About 280,000 babies, born 1991-2009, and their mothers, who are registered at GP practices in England, will be included. Duration of follow-up varies from a few months to over 20 years, depending on date of delivery. The main 'exposure' is ART, which comprises IVF, ICSI, and ovulation induction; other techniques (e.g. blastocyst transfer) will be examined, where possible. Outcome data include all consultations, diagnoses, prescriptions and referrals recorded by GPs, supplemented by data on hospital episodes and mental health treatment. Key outcomes for children include growth, illness and infection, atopic and allergic conditions, behaviour and developmental outcomes; and for mothers, physical and mental health. Multivariable regression models will be used to estimate adjusted rate ratios and odds ratios as appropriate, comparing outcomes in the ART and non-ART groups. An economic analysis will compare differences in NHS expenditure on the long-term provision of care for ART and non-ART pairs, and model future costs.

Concern among researchers that low consent rates since 2009 adveresly affect the representativeness of the HFEA dataset, means that the newest data are under-utilised. Methodological work to explore the impact of missingness, and to develop methods to reduce the potential bias, will be conducted. Qualitative research, exploring patients' decisions to consent (or not) to studies of ART will also be completed, through face-to-face interviews and thematic analysis.

This study will contribute to the evidence for the safety and costs of ART in England, and provide tools for future work using these rich data sources.

The project will produce the following outputs and impacts:

1. We aim to make the linked dataset available to other researchers. This would provide a valuable resource for ART research in the UK population, and the 'modern' data (post 2009) has the potential to form an on-going cohort of ART mother-baby pairs through continued CPRD data collection, if agreements can be reached with HFEA.
2. The analysis described here would be reported at suitable conferences and in open access, peer-reviewed journal articles describing the observed effects of ART on subsequent child health and development in this population. (Benefit: the research community, and clinicians who treat infertile couples by providing more evidence of long term effects of ART).
3. Similarly, the findings on the effects of ART on subsequent maternal health and wellbeing would be submitted to open access, peer-reviewed publication. (Benefit: the research community, and clinicians who treat infertile couples by providing more evidence of long term effects of ART).
4. There is some suggestion that researchers are not using the 'modern' (post 2009) HFEA data because of concerns over low participation rates, lack of representativeness and potential bias (personal communication, HFEA). We would provide evidence on the extent of bias in the HFEA dataset collected after the change to consent rules in 2009, and provide methods (such as weights) that can be employed to reduce the effects, and thus render the dataset more reliable. (Benefit: HFEA and research community, plus users of research findings who can be more confident in the results derived using methods to account for low consent rates).
5. We would provide recommendations for improving the consent rate for data to be used in research (Benefit: research community, ART clinics seeking consent and HFEA)
6. A journal article assessing the long-term cost to the NHS of additional use of primary care by ART mothers & babies would be submitted for publication (Benefit: NHS, those planning primary care costs)
7. Finally, a lay summary of the key findings will be provided. (Benefit: the general public (including women and families who will have contributed data to the study); those providing support for couples experiencing infertility such as Infertility UK; and those supporting families following ART such as ACeBabes).

In addition, the staff involved in this study would develop further research and professional skills that would benefit both themselves, and future employers. As the fellow, CC would benefit most from the training and development plans outlined here. However, there will be other benefits, such as employing a new Health Economist who will gain valuable experience of analysis of a large dataset. It is hoped that new collaborations proposed as part of this fellowship will foster further collaborative work.

The time frame for these outputs is within the 6.25 years of the fellowship funding. The impacts, in terms of providing resources for future use and adding to the body of evidence regarding the safety of ART, will be longer lasting. If the linked data can form the basis for a continuing cohort study, there could be on-going outputs from this data linkage project exploring the effects of ART on mothers and children over the coming years. In addition, the methodological work outlined here (Objective 5) would ensure that the data held by the HFEA could be considered robust. The weights we generate to compensate for low response rates could be provided for analyses by other researchers and would encourage scientists to confidently use the data, so that this resource could be more widely used both at the end of the project and in future. Please see also the 'Pathways to Impact' document supplied in the attachments.