The use of hospital data has significantly expanded in the past years to better understand treatment patterns and patients’ characteristics. Comparing treatment and outcomes between hospitals is an important method of ensuring good standards of care.
Undertaking studies that compare between hospitals, particularly between countries, remains a significant challenge due to data privacy concerns. “Federated” analysis, where hospitals can work together to complete the same analysis and share aggregate results while data never leaves the hospitals’ servers, allows robust multi-centre data-analysis to be undertaken without any need for data sharing.
In an effort to facilitate the above process, the OHDSI network has developed the Observational Medical Outcomes Partnership (OMOP) common data model (CDM). Once hospitals have formatted their data to this common, standardised CDM, an algorithm can be run by each hospital and share the report with other sites. This means than no individual patient data is leaving the hospitals, and the reports contain summaries only which means that the patients cannot be identified. The results are additionally checked by each sites’ disclosure guidelines before being shared.
This project is led from the UK by the National Institute for Care Excellence (NICE). It is looking at the example of how cancer survival differs between different European hospitals. Participation will allow us, NHS Lothian, as part of the DataLoch initiative to test our recently mapped breast cancer data and test our local processes needed to support federated analysis in a safe manner.
