My organisation is in the early stages of converting some of our anonymised EHR data into the OMOP format.
We have multiple versions of our most granular location values for an individual’s data - these areas’ boundaries are redrawn for every census, so a care site or person’s location may be in a different area depending on if you use the 2001, 2011 or 2021 boundaries.
We are keen to be able to record these different values for a care site or person’s location, as it is useful for future use in research once the OMOP process is complete, and for this niche situation we have not yet been able to figure out a good-practice way to do such a thing.
Some bad-practice ways to do this that we have considered:
-Using the unused address fields in the location table to store the different versions
-Breaking the 1-many relationship, and somehow having multiple locations for one person/care_site
I’ve tried to abstract details away from this problem to keep this post concise, but I may have done so too much; I’m very happy to share more information if I’ve made no sense. Thank you!
Once upon a time, there was a v6.0 OMOP CDM which contained the LOCATION_HISTORY table. This is exactly what you need.
Someone else asked this question on the forums and was pointed to the GIS WG as they are working on a similar issue. I suggest you reach out @kzollove and the GIS WG with your use case. I implore you to NOT modify the definition of any tables or fields in order to make your data fit. This will create a multitude of unforeseen and damaging problems downstream.
We will be following the current CDM; it is a shame that there isn’t a mechanism for location history in the current version, as we have good data for the location history of individuals.