We (Chris Brandt, Allison Zwingen both at UC Davis, and Joe Strecker at CSU) have been meeting twice monthly on these CDM issues. Chris/Allison has proposed implementing support for breed and species in two places within our version of the OMOP CDM - based on version 5.2 with the following modifications:
We added species and breed fields in the person table, along with foreign keys to a new owner table. This keeps patient demographics contained in one table instead of creating a new Animal table. Obviously ATLAS and other OHDSI tools can’t see these fields. However, our veterinary specific applications can use these fields without having to include #2 below to simplify SQL construction.
We added species and breed in the observation table not tied to a visit_occurrence_id. The observation_concept_id are SNOMED concepts already in the standard vocabulary. So ATLAS and other existing OHDSI tools can use this information, thus providing a backwards compatibility. One secondary benefit of storing breed information in the observation table is we can support multiple-breed patients. For example a EHR documented “German Sheppard/Cross” can register two concepts in the observation table - one for “German Sheppard” and a second “Mixed breed dog”. This use case came up when we wanted to only include pure breeds in a query.
The above is going into production at Tufts Cummings Veterinary School next week. We have over 340,000 patients and over 816,000 visits.
I’m also in the process of integrating ongoing research data - case reports (surveys) from REDCap - into our CDM as part of our expanding implementation of OMOP CDM to support veterinary research.
Are you a member of the CTSI OneHealth Alliance (COHA)?