Tessa May - Technical Consultant

The Observation Medical Outcomes Partnership (OMOP) Common Data Model (CDM) is gaining recognition in the UK as a powerful tool for transforming real-world healthcare data into standardised, research-ready formats, where either minimal or no data transformation is required to perform analytical research across data sets. The NHS Research Secure Data Environment (SDE) Network recently announced the adoption of the OMOP CDM as the standard for SDEs across England (for further details, please see the article recently published by HDR UK and authored by Dr Chris Russell, Head of Delivery, Data for Research and Development at NHS England).

By harmonising disparate data sources, OMOP facilitates efficiency, collaboration and interoperability in research. While there are many benefits to implementing the OMOP CDM, in this post we’ll consider some of the key challenges in doing so.

Not sure what the OMOP CDM is? Please see our previous post!

Interpreting the Specification

The OMOP CDM is an international specification and is designed to be flexible, however this leads to various ways in which the guidance can be implemented, especially when aligning field definitions to UK healthcare data. For example, fields for Person demographic data, such as gender, race and ethnicity have specific context and connotations, which may lead to inaccurate assumptions. Fields and tables may have uses across multiple domains. There are key concepts, such as the Observation Period, which are open to interpretation. Mapping requires collaboration across clinical, data and academic research experts to ensure it is accurate and that loss of context and data fidelity are minimised.

Mapping to a Standardised Terminology

A significant complexity in implementing the OMOP model is the standardisation and mapping of clinical terminologies and coding systems. The UK utilises different various coding systems, such as SNOMED CT, Read, LOINC, ICD-10, or custom concepts. These variations make it difficult to ensure consistency and comparability across different data sources, and the OMOP model requires that codes are mapped to one of the Standardised Vocabularies. Collaborative efforts involving clinical and informatics expertise are required to reduce the risk of inaccurate mappings.

Combining Disparate Data Sources

Another challenge, which can be considered somewhat unique to the UK, is defining mappings consistently across the complex, diverse and fragmented landscape of healthcare data. Data is scattered across multiple systems, and stored in different formats in each. Addressing the challenges described above must be repeated for each system, and may also be repeated for the same systems across organisations.

Governance and Stakeholder Engagement

Establishing effective governance structures and fostering stakeholder engagement are critical for successful implementation of the OMOP model. Bringing together policymakers, researchers, clinicians, and data custodians is crucial for developing guidelines, defining mappings, ensuring data quality, and addressing considerations regarding privacy and data protection. Furthermore, it is imperative that mapping definitions and implementations are maintained, which can be complex to govern across multiple systems.

So, how can you overcome these challenges?

Any data mapping exercise is complex and requires careful consideration, and mapping to the OMOP model is no different. A collaborative, systematic approach, with robust documentation and governance can increase efficiency, accuracy and consistency. Without this, the realisation of benefits resulting from attempting to standardise data will be limited.

Furthermore, encouraging data sharing initiatives and fostering partnerships can lead to a more unified and accessible data infrastructure. Here at Answer Digital, we are developing a systematic approach to implementing the OMOP model, which we will be sharing details of in our next post. We are keen to collaborate across communities, so if you would like to hear more, or are already using the OMOP model then please get in touch, and stay tuned for updates on our work!