OpenPREGnosis: Developing an Open Algorithm to Identify Pregnancy Episodes and Outcomes in National Electronic Health Records (EHRs): A Comprehensive Methodological Approach

Research Questions: Can an open-source algorithm be developed to identify unique pregnancy episodes in de-identified health records?

Can we integrate this algorithm as a tool within the national OpenSAFELY platform to advance maternal and fetal health research in England? Working with public and professional stakeholders what are the priority research questions for testing the algorithm?

Background: Epidemiology studies using electronic health records determine patterns of health and disease in specific populations and help define approaches to improve health outcomes. The new OpenSAFELY secure analytics platform enables research across 58 million patient records in near real-time.

However, studies improving maternity care are challenging, as healthcare is delivered across different NHS sites and each pregnancy is unique in length and outcome. This makes it difficult to accurately identify and characterise pregnancy episodes.

Aim: To harness the full potential of OpenSAFELY by developing a robust method - an open-source algorithm- to accurately identify unique pregnancy episodes and outcomes, empowering health research to improving outcomes for mothers and babies.

Objectives: 1) Conduct a rapid literature review to consolidate and critically evaluate current methods, forming the basis for algorithm development, 2) Incorporate iterative data-driven approaches in the development and validation of the algorithm - OpenPREGnosis, 3) Consensus building with clinical stakeholders to guide critical algorithm decisions, 4) Campaign for standardisation in recording clinical events 5) Iterative validation of the algorithm, 6) Adopt a patient-centred approach engaging with the public throughout, 7) Outline requirements for implementation within NHS Secure Data Environments (SDEs).

Methods: Develop an open-source algorithm within the OpenSAFELY platform, incorporating information from published literature, professional stakeholders, and national statistics for validation.

Work with public contributories to co-develop a national survey and co-deliver public workshops to define patient priorities, and direct focus of pilot research studies and NHS implementation.

Timeline: Months 1-4: Conduct a rapid literature review, stakeholder recruitment, survey development and introduction workshops. Months 2-21: Conduct iterative algorithm development and validation cycles, building stakeholder consensus. Months 5-10: Conduct a national survey and prioritisation workshops.

Months 13-23: Hold additional workshop to co-prioritise and implement pilot studies to assess the algorithms performance.

Months 19-24: Disseminate findings through journals, conferences, and public channels, with best practice campaigns and develop funding applications, and work with NHS SDEs.

Impact and Dissemination: Using the UK s largest platform of linked electronic health records (OpenSAFELY), OpenPREGnosis will revolutionise pregnancy research.

The development and testing of the open algorithm within OpenSAFELY allows scientific re-use and rigor, whilst minimising research waste.

The open and reproducible nature of our algorithm will allow rapid identification of pregnancy cohorts in research and NHS SDE databases, leading to better quality, larger-scale, reproducible pregnancy research in England and NHS implementation.

Researchers and healthcare professionals will be able to use OpenPREGnosis to rapidly assess the impact of interventions.

Working with the public to raise awareness of de-identified health data use for research, this collaborative model allows patients to contribute to the development of future research questions aligned with genuine patient concerns, for improved outcomes in maternal and fetal health.