Standardised Data: The OMOP Common Data Model
Current analytical process involves several databases for which data may duplicate. Joining such dataset to allow data mining is a significant, labour-intensive task, where manual curation against electronic health records is often necessary and time consuming. Databases may also use slightly different definitions for similar variables, making the process even more challenging to avoid inconsistencies.
This becomes exponentially more complex as collaborations increase. One simple way to counter this is to ensure that all hospital use the same format for each variable and use the same standard definitions and code list.
The OHDSI community is offering a standard common data model called OMOP (Observational Medical Outcomes Partnership), for all participating hospital to use. This would facilitate collaborations, increase sample size and consistency across sites, and ultimately benefit patient care.
The OHDSI community has a world-wide ambition, and pilot projects have showed significant promise involving large scale investigations.
The OMOP common data model has now extended to cancer data, and the Edinburgh Cancer Informatics is very excited to be part of this network.
Figure 1: the process of standardising data – a common language across all hospitals (source)
The Edinburgh Cancer Informatics successfully mapped data from the SESCD database for early breast cancer as a test project to the OMOP model, using funding from the EHDEN community, a partner of OHDSI. Future work will involve adding data from the cancer registry, to allow us in participating in a cancer survival study organised by OHDSI. Please find additional information about the Edinburgh Cancer Informatics mapping here.
Figure 2: The OMOP mapping (source)
One additional benefit of this OMOP model, is that all patient data stays on the local servers, and is not made available to any sites/organisation participating in the study as it adopts a federated network approach. This means that the patient data is not leaving the hospital, and is not made available to external partners to ensure patients anonymisation. Aggregated data will be used as output, and is subjective to NHS disclosure policy.
Please visit the Edinburgh Cancer Informatics portal entry here for additional information.
Spyro Nita, Architect (s.nita@epcc.ed.ac.uk)
Nicola Symmers, Developer (nicola.symmers@ed.ac.uk)
Mahéva Vallet, Senior Research Manager (maheva.vallet@ed.ac.uk)
Colin McLean, Senior Research Fellow (colin.d.mclean@ed.ac.uk)
Peter Hall, Medical Oncologist and Clinical Lead of Edinburgh Cancer Informatics (p.s.hall@ed.ac.uk)
