Hi folks,
Apologies if this has been asked before but I can’t find anything in the forums about this.
I’m a Data Manager at Genomics England. A couple of years ago, we spent some time mapping the curated data sets we get from the NHS to the OMOP CDM. Our latest set of mappings is here:
These are PDFs at the moment, but we’ll be looking to add some machine-readable maps, probbaly early next year. In the meantime, there’s definitely some room for improvement with them. Having to work with curated data rather than doing a direct ETL from a hospital EHR system has left us with with some gaps that we can’t easily fill. For example, having to work with the rolled-up drug groups in systemic anti-cancer therapy instead of detailed regimen information means that the drug-era and drug-exposure data is pretty poor. There were some data points - usually lists of allowed values published by the NHS - whose meanings didn’t translate directly to an OMOP concept, standard or otherwise. We also ended up making an executive decision to exclude all organisation data unless the place in question provided direct patient care or was the patient’s registered GP, as otherwise the care_site table would have filled up with unwanted data.
I’m sure we can’t be the only company running a TRE that has encountered this problem, and I’d be keen to hear from and ideally collaborate with anyone else who has encountered this.
I’d be most grateful for your thoughts.