Transforming Pregnancy Care with Data
A WRH Research Group harnessing clinical data to predict and improve pregnancy outcomes through innovative research and technology.
Oxford Growth Restriction Identification Programme (OxGRIP)
About the research study
OxGRIP is a research group comprising of clinical academics and clinicians led by Prof. Lawrence Impey who are using routinely collected clinical data on pregnancies within Oxford University Hospitals NHS Foundation Trust to identify risk factors for a poor pregnancy outcome in order to guide new pathways of care.
How THE study can help
Our initial aim of OxGRIP was to investigate the practical, structured use of ultrasound in reducing pregnancy risk.
This aim has been extended to reduce adverse pregnancy outcomes through smarter use of data to more accurately predict pregnancy risk.
Through this we hope to:
1) Identify pregnancies where intervention could be beneficial.
2) Reduce intervention where intervention wouldn't be beneficial.
3) Promote equality in pregnancy outcomes.
We are using conventional statistical methodology as well as machine learning in collaboration with other groups. This is a joint project between Oxford University Hospitals NHS Trust and Oxford Deep Medicine and are funded by Social Finance.
Why the study is so important
Nearly 1 in 100 babies die during or soon after pregnancy, even where healthcare systems are advanced. This rate is now near static and particularly high in marginalised groups and ethnic minorities.
Aside from the human cost, adverse events in maternity provision are expensive. In the UK they account for about half of the entire NHS litigation budget and cost more than the provision of the care itself. As attempts are made to reduce stillbirth nationally, maternity care is receiving considerable, frequently negative, press in the UK and abroad.
The Importance of Predicting At-Risk Pregnancies
Most pregnancies proceed without the need for medical intervention, making it crucial to predict those at risk. Accurate prediction enables the targeting of resources and interventions to those in need, while preventing unnecessary and potentially harmful interventions for those at lower risk.
Despite established risk factors, many maternal deaths occur in women not currently deemed high risk. This has led to an increased medicalisation of all pregnancies, with a trend towards earlier births and surgeries, which can cause significant and lifelong harm, including early births (iatrogenic for the baby) and psychological and physical trauma for the mother.
Limitations of Current Risk Prediction Methods
Current risk prediction methods largely rely on a checklist approach with binary values and no weighting of risk factors.
This qualitative method results in inaccurate risk estimation and a high rate of false positives. Some early pregnancy screening tests (such as the QUIPP and Tommy’s apps) are gaining traction, but they provide risk assessments only in early pregnancy and cannot be updated with new data or pregnancy events.
The accuracy of quantified risk prediction up to 34 weeks is high. - Prof. Lawrence Impey
Developing a Staged Risk Prediction Model
Our goal is to create a staged risk prediction model for adverse pregnancy outcomes, which will be updated with both prior risk data and new clinical findings at key pregnancy stages: 20 weeks, 28-30 weeks, 36 weeks, and immediately before labour. Unlike current practices, our model will be quantitative and staged, offering more accurate and timely risk predictions.
Focus on Predicting Placental Function Impairment
While many stillbirths remain unexplained, impaired placental function likely accounts for over 50% of cases and has long-term health consequences. Our initial project focus will be on predicting impaired placental function. In collaboration with the Oxford AHSN (now Health Innovation Thames Valley), OUHFT introduced a care pathway in May 2016, offering universal uterine artery Doppler at 20 weeks and a universal growth restriction screening test at 36 weeks.
The predictive value of the 20-week Doppler has been demonstrated, and we have collected high-quality data that will be used for our risk prediction model.
The 3 principal time periods where risk prediction will be applied
1. Prediction of severe preterm (<34 weeks) pregnancy loss.
This will focus on prediction of growth restriction
2. Prediction of later (>34 weeks) pregnancy loss
This will focus on growth restriction and later pregnancy risk factors
3. Prediction of severe neonatal illness during birth
This will involve collaboration with AI- based intrapartum monitoring methods in development and will provide prior risk data to the attentiveCTGnet deep learning model being developed in collaboration between The University of Oxford and University of Portsmouth.
Enhancing Usability and Integration
For the risk prediction tool to be usable, it must be compatible with automated extraction of individual risk factors from EPR systems, allowing for easy, touch-of-a-button risk prediction. This tool will combine risk prediction with clinical advice and facilitate shared decision-making.
UPDATE: What we have learned so far
- The accuracy of quantified risk prediction up to 34 weeks is high.
- Universal uterine artery Doppler assessment, either at 12 or at 20 weeks, is probably essential to risk prediction before 34 weeks
- Beyond 34 weeks it is more complex and is likely to perform less well
- Universal late pregnancy ultrasound for fetal growth restriction may reduce adverse outcomes without necessarily greatly increasing intervention
- Ultrasound accuracy requires improvement and AI-based methods show promise
Useful links
Deep Medicine
The Deep Medicine programme combines pioneering data analytic approaches, deep and machine learning techniques, and interdisciplinary collaboration to generate insights that will help tackle some of the major causes of death and disability worldwide.
Oxford Labour Monitoring
The group is committed to preventing injury of babies during labour and delivery, caused by lack of oxygen in utero. Our work will benefit families, clinicians and healthcare systems.
Latest news
For all the latest news from the Nuffield Department of Women's & Reproductive Health, please visit of blog page.
Latest Publications
Ultrasound Obstet Gynecol. 2024 Apr 26. doi: 10.1002/uog.27668. Epub ahead of print. PMID: 38669595. Dockree S, Aye C, Ioannou C, Cavallaro A, Black R, Impey L
Ultrasound Obstet Gynecol. 2024 Jan;63(1):98-104. doi: 10.1002/uog.26305. PMID: 37428957. Robertson K, Vieira M, Impey L.
J Matern Fetal Neonatal Med. 2023 Dec;36(1):2152670. doi: 10.1080/14767058.2022.2152670. Epub 2022 Dec 8. PMID: 36482725.Mathewlynn S, Beriwal S, Ioannou C, Cavallaro A, Impey L.
BJOG. 2023 Jun;130(7):791-802. doi: 10.1111/1471-0528.17395. Epub 2023 Feb 13. PMID: 36660877. Aderoba AK, Ioannou C, Kurinczuk JJ, Quigley MA, Cavallaro A, Impey L
2022 Sep;60(3):373-380. doi: 10.1002/uog.24971. PMID: 35708532. Mathewlynn S, Impey L, Ioannou C.
BMJ Open. 2022 Mar 23;12(3):e058293. doi: 10.1136/bmjopen-2021-058293. PMID: 35321896; PMCID: PMC8943771. Aderoba AK, Nasir N, Quigley M, Impey L, Rivero-Arias O, Kurinczuk JJ.
J Clin Ultrasound. 2021 Jun;49(5):442-450. doi: 10.1002/jcu.23014. Epub 2021 Apr 6. PMID: 33822384. Procas-Ramon B, Hierro-Espinosa C, Salim I, Impey L, Ioannou C.
PLoS Med. 2021 Jan 15;18(1):e1003503. doi: 10.1371/journal.pmed.1003503. Erratum in: PLoS Med. 2021 Apr 22;18(4):e1003613. PMID: 33449926; PMCID: PMC7810318. Salim I, Staines-Urias E, Mathewlynn S, Drukker L, Vatish M, Impey L.
Principal Research Team
-
Lawrence Impey
Visiting Professor in Fetal Medicine
Oxford University Hospital staff:
Mr Angelo Cavallaro
Dr Sam Dockree
Mr Kashif Qureshi
Dr Mike Shea
-
Rebecca Black
Honorary Senior Clinical Lecturer
-
Sam Mathewlynn
Clinical Research Training Fellow & DPhil Student
Other members of the team
Dr Christos Ioannou
Dr Katherine Robertson
Dr Marta Garbagnati
Visiting research fellows
Dr Elena Dalberti
Dr Chiara Granieri
Dr Leh Sii Ping
PPI lead
Ms Joy Randolph
Professor Kazem Rahimi
Dr Shishir Rao
Collaborators
Growth trajectories and EFW estimation
Prof. Alison Noble
Miss Emily Peacock (D Phil student)
Prof. Basky Thilaganathan
Prof. Ivan Jordanov
Prof. Antoniya Georgieva
How can you help?
You can support the ongoing work of the OxGRIP project through donations, collaborations and research support. If you wish to support our work, please contact us or email Dr Christina Aye using the button below.