Inequalities in stillbirth and preterm birth and their risk factors

The University of Bristol (UoB) requires data from the Maternity Services Data Set (MSDS), for the purpose of investigating inequalities in rates of stillbirth (SB) and preterm birth (PTB) and their risk factors in England.

Background:

The Secretary of State for Health has defined a national target to halve SB by 2025, with a reduction from 4.7 per 1000 to 2.3 per 1000. There is a similar ambition to reduce PTB from 8% to 6% in the same time frame. There has been an increased national focus on improving maternity outcomes, with a range of initiatives developing from the government, the NHS and professional bodies. Though much is still to be discovered, we know a lot about excellence in maternity care, but less about how to ensure that this is accessible to all women beyond centres of excellence. The main purpose of this study is to contribute towards achieving the primary aim of the consortium to reduce SB and PTB in the UK.

The rates of SB have been mapped for 2017 by MBRRACE-UK: Mothers and Babies: Reducing Risk through Audits and Confidential Enquiries across the UK. The rate of SB in 2018 is 4.1/1,000 total births. Wide variations have been identified throughout the UK. Some NHS Trusts have SB rates of more than 15% lower than the national average, but other units have rates more than 5% higher than the national average. These inequalities remain after adjusting for differences in patients and Trust characteristics. Inequalities in PTB have been identified in England and Wales by maternal residential area (regions) but there is no data on variations between NHS Trusts and over time. The risk factors that contribute to these geographical variations remain unclear. Identifying and understanding the variations in SB and PTB should help care providers to identify best and poor practices and evaluate the effectiveness of intervention or guidelines aiming to improve care practices and clinical outcomes. However, clinicians and mothers need to be able to identify the risk of PTB and SB as early as possible to prevent these outcomes.

The literature on risk factors of PTB and SB is substantial. These risk factors could be used to form the basis of a predictive tool that would help clinicians identifying at-risk women at booking or during the early antenatal visit and therefore identify adequate antenatal care pathways for these mothers. The tool could also help identify women at risk of PTB and/or SB throughout their pregnancy, especially during acute hospital preterm presentation or late pregnancy presentation and offer them appropriate screening, care and delivery strategies.

By analysing data from the MSDS data, would provide the information necessary to address these knowledge gaps. It provides a standardised collection of data on maternity care, maternal and perinatal outcomes, across all of England since 2015. It also contains numerous maternal, babies and NHS Trust characteristics including clinical features. This is therefore a unique source to develop and test risk prediction tools for SB and PTB. The size and coverage of the MSDS data offer the statistical power to consider rare variables and rare outcomes such as SB. Moreover, by providing information from multiple maternity units, it captures the diversity of clinical practices across England and generalisability to the results. Finally, MSDS data are ideal to conduct quasi-experimental studies at the national- or local care provider-level investigating the change in SB and PTB induced by the implementation of new clinical guidelines, or public health initiatives generally implemented independently of the data collection process and the research evaluation team. The study team will also assess changes on PTB or SB rates over time and/or induced by specific intervention impact(s) on neonatal death (ND), i.e. baby death within 30 days of delivery.

The lawful basis for processing personal data under the UK GDPR is:

The study team's project is justified by Article 6(1)(e) of the GDPR "processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller". Data processing is needed to help describe evidence of inequalities and geographical variation in rates of PTB and SB across maternity units, and to identify risk factors that would explain why such variation exists.

The lawful basis for processing special category data under the UK GDPR is:

The study team's project is justified by article 9(2)(j) of the GDPR "processing is necessary for reasons of public interest in the area of public health, such as protecting against serious cross-border threats to health or ensuring high standards of quality and safety of health care and of medicinal products or medical devices, on the basis of Union or Member State law which provides for suitable and specific measures to safeguard the rights and freedoms of the data subject, in particular, professional secrecy". This is justified as this project's aim is to drive improvements in the quality and safety of maternity care and to improve outcomes for mothers and babies.

Cohort identification:

Data subjects are those in the MSDS data. All births which occurred in England and captured in the MSDS data since the inception of the dataset (2015/2016) to latest available will be eligible. All the non-preterm and non-stillbirths data is required to provide an appropriate cohort control for all the preterm and stillbirth in England captured in the MSDS database from 2015 – latest available.

Data required:

For the cohort described above, data is needed from the MSDS dataset. NHS England would provide a one-off report of MSDS data. The team are only requesting access to the pseudonymised variables from the MSDS data source. One of the main aims of this project is to develop and validate an algorithm/model for early detection of pregnancies at risk of stillbirth and/or preterm birth for prompt interventions. Thus, they are interested in estimating the absolute risk of having a stillbirth/preterm birth which will require the whole data rather than using a fraction of all the non-preterm and non-stillbirths data as control.

The outcomes of interest are stillbirth (SB) and preterm birth (PTB) and they will consider how changes on these birth outcomes have impacted on neonatal death (ND).

- SB is defined as the death of a baby before or during birth after 24 completed weeks of gestation. SB is determined by clinicians as part of the care delivered in the local maternity unit. The SB status is recorded by clinicians in the hospital electronic medical records and directly captured in the MSDS. ND is defined as death within the first 30 days of life.

- PTB is defined as the birth of a baby before 37 weeks of pregnancy (i.e. <37+0 weeks). PTB status is derived from the length of gestation at birth recorded by clinicians in the hospital electronic medical records. Gestational age at birth (age of baby at birth) is recorded in the MSDS and will be used to define the preterm birth status and the different level of prematurity.

Risk factors of interest will include mothers age, ethnicity, obstetric history (e.g. previous pregnancies, caesarean sections, preterm births, birth weight), antenatal care (e.g. body mass index, smoking, alcohol, diabetes), labour and delivery (e.g. induction, drugs in labour, delivery method).

To describe geographical variation in rates of PTB and SB outcomes, the team require information on the name of maternity units at which the delivery took place, including name of NHS Trust, and higher levels of geography including the clinical commissioning group (CCG) areas.

Regarding date and time of event variables, the MSDS contains information on time of admission, time of onset of labour, time of second stage, time of third stage, time of induction and time of delivery. The team would like to be provided with variables containing the duration of labour, duration between stages or duration between the time of a particular stage and time of admission. The derived variables (date as MM/CCYY, time and duration) will maintain patient confidentiality.

Data minimisation reasoning:

The study team are only requesting MSDS data relevant to assessing the outcomes of PTB and SB and have limited the data requested to the cohort described. The team are not requesting further hospital data beyond the birth outcome.

Data analysis plan:

1. Explore the inequalities in the rates of SB and PTB between care providers across England, their variations over time;

Rates of SB and PTB will be derived for each NHS maternity unit or Trust, depending on the availability of information at the unit level. A cross-sectional investigation of SB/PTB rates for a particular year will be supplemented by a longitudinal investigation with rates estimated for each care provider and by time-period of relevant length. SB is a rare outcome and to retain patient confidentiality, time-period will be defined to aggregate no less than five cases/patients per unit/Trust for any given period. Funnel plots, statistical process control charts, Geographical Information System maps and caterpillar plots will be considered to report the findings.

Multilevel generalized linear regression models will be used to compute the rates and their 95% Confidence Intervals. This will allow the team to determine whether relevant risk factors of interest explain observed geographical variation in rates of PTB and SB outcomes across maternity units.

2. Examine the impact of SB/PTB on neonatal deaths.

Several interventions including national guidelines, specific clinical or organizational interventions, or any other initiatives susceptible to impact adverse outcomes such as SB or PTB have been implemented. This work package will also examine how changes on the rates of SB and PTB have been affecting neonatal deaths. This work package will involve comparing trends in SB, PTB and ND.

Also, the effect of differences in the incidence of SB/PTB between maternities on neonatal death will be examined using generalised linear (mixed) regression model.

3. Investigate the impact of initiative(s) implemented to improve maternity care or SB/PTB;

We will identify initiatives aimed at improving maternity care provision and/or care quality over the time period of interest (2015 to present). These include national guidelines, specific clinical or organizational interventions, or any other initiatives susceptible to impact adverse outcomes such as SB or PTB. The initiative could be implemented at the national level, or at the local level within one or several maternity unit(s) or Trust(s). This work package is based on a quasi-experimental (before-after) study design. We will use the date of introduction of the studied initiative to define the relevant period of exposure: births that occurred before the intervention will be considered

as unexposed and births that occurred thereafter will be considered as exposed to this initiative Statistical models such as generalised linear (mixed) regression models, including interrupted-time series, will be used to perform these analyses.

4. Further the knowledge on the risk factors of SB and PTB prior and develop risk prediction models for these two adverse outcomes. We will investigate the associations between SB status and maternal, baby and care provider level characteristics available in the MSDS data. A similar approach will be conducted to investigate the risk factors of PTB. Two approaches will be used. We will use traditional generalised linear (mixed) regression such as logistic or Poisson models to identify the characteristics associated with the outcome of interest. These factors will be used to define predictive models for SB and/or PTB. The validation of the predictive model will be done with strategy such as the bootstrap approach or Area under the Curve. We will repeat these analyses with machine learning models using supervised learning approaches (Lasso regression, Support vector machines, Gradient boosting machines model etc) and explore the potential for unsupervised approaches to refine these predictive models.

Data Controller statement:

The University of Bristol is the sole data controller and processor, given the University of Bristol are solely determining, through the study team, the way the data is being processed. The University of Bristol is the data controller responsible for determining the purpose for all data flows described and will be the data controller for the data received.

Funding:

This study is funded by the charity TOMMY’s, which established a National Centre for Maternity Improvement in September 2019 aiming to identify the best care practices to reduce SB and PTB and spread them to all care providers across the UK. Tommy Centre is being led by the Royal College of Obstetrics and Gynaecologists (RCOG) with the Royal College of Midwives (RCM). The team at the University of Bristol is a member of the new TOMMY centre and is leading on statistical analysis of national maternity data to address the aims of the centre.

Under this agreement, data on all births that occurred in England and recorded in the MSDS database from April 2015 to latest available will be transferred to the University of Bristol by NHS England.

Regarding date and time of event to be checked, the team are requesting that the data provided to them have dates transformed to the MM/CCYY format to protect patient confidentiality. Time variables in the MSDS are generally registered as CCYY-MM-DDTHH:MM:SS. The team are requesting that all time variables are provided to them as HH:MM:SS without the DD/MM/CCYY. The team are also requesting for the time of events to be converted in duration as explained in the section above. The derived variables (date as MM/CCYY, time and duration) will maintain patient confidentiality. The data will be exclusively stored on the University of Bristol SafeHaven (server) and will not be accessible to any third parties.

Data flow:

1. A patient-level, pseudonymised MSDS dataset (no personal data are requested, and mums/babies will be identified with a pseudonymised ID) would be provided by NHS England.

2. NHS England will supply the pseudonymised data to the University of Bristol via Secure Electronic File Transfer (SEFT).

3. The University of Bristol will store data in the University of Bristol SafeHaven and access to the data will be restricted to the statistical team at the University of Bristol.

The Data will be accessed by authorised personnel via remote access.

The Controller(s) must confirm and provide evidence upon audit by NHS England that access via any remote device complies with the data security obligations within this DSA and the Data Sharing Framework Contract.

For remote access:

- Remote access will only be from secure locations situated within the territory of use (as further restricted elsewhere within the DSA if so done) stated within this DSA;

- Access controls granting users the minimum level of access required are in place;

- Remote access is only via secure connections (e.g., VPNs or secure protocols) to protect data;

- Multifactor authentication (MFA) is required for remote access;

- Device security, including up-to-date software and operating systems, antivirus software, and enabled firewalls are utilised for the remote access;

- All remote access is undertaken within the scope of the organisation’s DSPT (or other security arrangements as per this DSA) and complies with the organisation’s remote access policy.

The above applies in addition to any condition set out elsewhere within the DSA (e.g. who may carry out processing, and for what purpose).

Statistical analyses of the dataset will be performed as described in section 5a. All outputs will contain only data that is aggregated with small numbers suppressed.

The MSDS data will be located at the University of Bristol SafeHaven, which has established procedures for the transfer of data, its secure storage and for erasing data at the end of projects (www.bristol.ac.uk/infosec/policies/docs/). The data will be access-controlled at the University of Bristol in safe data havens, with password and firewall protection that guards against external users and access will be limited to the staff named in this application.

Only substantive employees of the University of Bristol will access the disseminated data and only for the purposes described in this agreement. Specified University of Bristol study personnel will be granted access to the data safe-haven after they have completed the necessary information governance training; The Project Lead will have responsibility for managing those individuals who have access. The data will be managed by a member of the project team based at the Learning and Research Building at the University of Bristol. The data will be used exclusively for the purpose of this project.

Once this agreement is near to expiring an extension application will be submitted to NHS England, to allow the continued processing and storing of the data. As at the end of the study, the data will need to be safely held in a password protected project-specific safe data haven at the University of Bristol for 42 months and, in that time, it will be accessed only to answer questions arising from publications and other publicity. The expected time frame for completion of the data processing, production and dissemination of the outputs is currently 4 years, with a further 18 months retention of data after this to respond to changes based on peer review comments from journals and from funding bodies.

The study team are requesting a pseudonymised dataset and there will be no requirement/attempt to re-identify individuals. No data will be shared.

The NHS England data will not be linked to any other data.

This Data Sharing Agreement does not permit data to be made available to any third parties except in the form of aggregated outputs with small numbers suppressed in line with the HES Analysis Guide unless a third party has received separate approval from NHS England to receive the data covered by this Agreement. To prevent inadvertent reidentification of the participants, UoB plan to take the following steps:

(1) Only applying to access depersonalised, pseudonymised MSDS data

(2) No personalised data are requested and mothers/babies will be identified with an anonymised ID

(3) The data will be securely stored in a University of Bristol server and access to the data will be restricted to the statistical team at the University of Bristol and a PhD student employed by the University of Bristol

(4) The study team plan to merge data from multiple years before analysis to prevent the results from having small numbers: the statistics will be reported in output using rates, .ie. aggregated numerators and denominators, that will not allow identification of individual data or individuals

(5) If these rates are derived from small data, i.e. small numerators <5 cases, we will consider, whenever possible to elongate the length of each period modelled (i.e. 3 months instead of one month, 6 months instead of 3 months etc) to prevent the production of rates based of data too sparse.

(6) If despite these measures, results include small rates that are still based on numerators, or numbers <5 cases, they will be suppressed and not reported to secure patient confidentiality

Throughout all stages of this programme, the key stakeholders including the patients, public, the Royal College of Obstetricians and Gynaecologists, the Royal College of Midwives, NHS managers and healthcare professionals will be engaged for interpretation, dissemination, and direct communication of the main findings.

The data obtained from NHS England will be examined to generate the following publications to achieve the project's objectives.

(1) Inequalities in the rates of SB between care providers/LSOA across England

(2) Inequalities in the rates of PTB between care providers/LSOA across England

(3) Impact of the implemented interventions to improve SB.

(4) Impact of the implemented interventions to improve PTB.

(5) Associated risk factors of SB.

(6) Associated risk factors of PTB.

(7) Development and validation of a prediction model for SB.

(8) Development and validation of a prediction model for PTB.

Some of the journals where the team plan to publish these papers include but are not limited to the following: the Lancet, BMJ, Journal of the American Medical Association, British Journal of Obstetrics and Gynaecology and European Journal of Obstetrics & Gynaecology and Reproductive Biology and American Journal of Obstetrics and Gynaecology.

All outputs will adhere to the NHS England guidelines so that data are only shown in aggregate form with small numbers suppressed.

The research team will work alongside Tommy’s, other charities and learned societies to disseminate the findings of this study using established platforms that include social media such as Twitter and a study website, as more pregnant women are now turning to these resources for information about pregnancy. The research team will develop plain English summaries of findings for communication to women of reproductive age and members of the public.

The data produced by this study will inform NHS England, for whom elimination of inequalities and reduction in SB and PTB are priority targets. Incorporation of the findings from this project to maternal care services will not be challenging as the study will be led by the Royal College of Obstetricians and Gynaecologists and the Royal College of Midwives.

Disseminating outputs for pregnant women and the public:

The outputs of this research will be hugely valuable in creating messages and information products directed at very large groups including pregnant women, women of reproductive age, men, the wider public. Several social media platform (Twitter, Instagram, Facebook) will be deployed to disseminate the findings of this study to pregnant women and the public.

Disseminating outputs to collaborators and stakeholders:

Tommy’s research centre is supported by representatives from the main charities which work to reduce stillbirth, miscarriage and preterm birth such as SANDS. The centre is also supported by the RCOG Women’s Voices Involvement Panel, an online group of over 500 members of the public who want to use their experience of women’s health services to influence the work of the College and the wider women’s health sector. A member of the RCOG Women’s Network is also involved in the centre’s advisory group. Women have therefore been involved in the writing of our grant application and are involved in the development of each work package. They have contributed to the development of the above research questions and will be involved in the interpretation and presentation of the related findings. The research team will continue to work with collaborators and stakeholders to enable the wide dissemination of findings.

Disseminating outputs to commissioners, operational managers and facilitators of change:

The team plan to engage the CCGs to get their buy-in to adopt the findings from this study. This study has the potential to contribute towards reducing SB and PTB by understanding the differences in SB/PTB, assessing the impact of the implemented interventions, evaluating the impact of changes on SB/PTB on ND and developing a clinical decision-making tool for early identification of pregnancies at risk of SB/PTB for prompt interventions. The research group will adopt multiple approaches such as workshops, policy to dissemination, led by the Royal College of Obstetricians and Gynaecologists and the Royal College of Midwives which have skill and capacity in this area.

Disseminating outputs to health care professionals:

The TOMMYs national centre is being led by the Royal College of Obstetricians and Gynaecologists and the Royal College of Midwives. Other relevant healthcare professionals were involved in the development of this study. Continuous engagement of all relevant healthcare professionals in pregnancy care will be sustained throughout the implementation of the study through meetings/newsletter. The Royal College of Obstetricians and Gynaecologists and the Royal College of Midwives will lead the dissemination of the study’s findings to their colleagues through various platforms including scientific meetings, national conference and workshops.

As this study is support of a PhD Research study a PhD Thesis will be produced and published.

This project is a core part of the work planned by our Tommy’s centre. This centre is contributing to the Department of Health target to half stillbirth (SB) and preterm birth (PTB) by 2025. The quantitative evidence provided by the proposed work will help to describe and understand the inequalities between care providers. This will help to identify units with low rates of SB and PTB, and those requiring support. The proposed work will also help to identify the effectiveness of existing national or regional/local programmes and provide evidence around the sustainability of their effect or relevance of scaling up a regional initiative to the national level. Finally, our work on the prognostic model should inform the development of clinical tools which would help clinician identifying at-risk patients as early as possible, to review individualised and tailored care.

The study’s overarching aim is to contribute towards reducing the incidence of SB and PTB by identifying high-risk pregnancies for prompt interventions. By identifying cases of high-risk pregnancies to receive the currently recommended interventions for the management of PTB and SB in the UK would assist in alleviating the current burden of stillbirths and preterm births nationally.

There is wide variation in standardised stillbirth and perinatal mortality rates throughout the UK with over a 20% difference between the highest and lowest reported rates whether by geographical region or Health Trust. This difference is not explained by a lack of current research recommendations, but rather by variation in local adoption and implementation of existing guidelines and good clinical practice, along with differences in availability and organisation of local resources. This study will explore in detail the disparities in SB/PTB, their range, potential associations/causes and provide recommendations. Findings from this study will also reveal the impact of SB/and/or PTB on ND. These findings will help understand and develop clinical decision-making tools to personalise care for women.

The study will support a PhD research study.

As part of our overall project on stillbirth(SB) and preterm birth(PTB), our research team has investigated the change over time in SB using ONS data (https://pubmed.ncbi.nlm.nih.gov/36112509/), the variation in the rate of SB and PTB using the National Maternity Perinatal Audit (NMPA-HQIP) registry (under review) and is now using this NHS data registry to investigate the risk factors of SB and PTB across all NHS England Trusts among nulliparous person (person who has never given birth to a live baby). There is extremely little research on risk factors of SB and PTB in nulliparous women, representing around 52% of pregnant women in England. Most of the available evidences on risk factors focus on women with a previous history of pregnancy, restraining the evidence available to the NHS and worldwide clinical communities, by putting away around half of the at-risk populations on which clinicians have very limited information to base their clinical judgement.

Current results, undergoing internal revisions, before submission to peer-review journals show that:

-For SB: over 870,000 nulliparous women were under the care of the NHS England between 2015 and 2019, and 2,568 stillbirths occurred. Maternal modifiable characteristics included, elevated body mass index, smoking, substance misuse, lack of folic acid consumption, antenatal booking after 13 weeks, and fetal growth restriction. We were unable to develop a clinically prediction tool as the area under the curve showed a very low predictive ability, indicating the need for further research in the nulliparous population.

-For PTB: over 870,000 nulliparous women were under the care of the NHS England between 2015 and 2019, and 56,353 preterm births (6.64 preterm births per 100 births) occurred. Of which, 49,206 preterm births were classified as spontaneous (5.63 per 100 births) and 6,701 were classified as iatrogenic (0.77 per 100 births). In early pregnancy (~8 to 12 weeks gestation) overlapping modifiable risk factors for spontaneous and iatrogenic preterm birth included, elevated body mass index, substance abuse, smoking, late booking

>13 weeks. Non-modifiable early pregnancy risk factors included older maternal age, previous miscarriage, conditions at booking, pre-existing diabetes, mental health condition at booking, family history of hypertensive disorder, index of multiple deprivation and complex social factors. Independent risk factors for spontaneous preterm birth included ethnicity, low body mass and a family history of pregnancy hypertension. The risk of iatrogenic preterm birth increased with increasing BMI and age.