I am Dagmar from the Amsterdam UMC hospital. I am implementing OMOP-CDM to provide researchers from clinical data from our hospital.
I have a question around laboratory measurements. I have patients with laboratory measurements taken on the same moment, with the same results. However, this was a different sample taken on a different location. This creates duplicate rows in the measurement table.
I was thinking about adding a collumn with a sample ID in it, so that our researchers can see that this is not a duplicate row but a second sample with the same results. Is this the correct solution or are there betters ways to solve this issue?
Thank you for your help, I really appreciate this forum!
Yeah, this is pretty typical. You actually don’t need both records. You could skip the one when it was taken and just leave the one when you got the result. Of course, that could be off by a day, but for most use cases this is totally fine.
If you do want to keep both you should distinguish them by the measurement_type_concept_id. You could use EHR order for the one when the blood was drawn and the test was order and Lab for the record from the lab.
You can add columns as well, but of course no researcher other than yourself will be expecting such a thing. It is off-standard as far as the OMOP CDM is concerned, which means, for all intents and purposes it doesn’t exist.
So, you have two lab measurements, but they look identical since you don’t bring over the location (assuming this is a body part) information. I see three options:
You implement the Specimen table which contains information about anatomic site
You leave them both since they are two different measurements. And in your end user documentation, you make note of this.
You leave them both since they are two different measurements. And add an extension column with the sample ID, but only you will know it’s there. All the OHDSI tools and network research use the canonical CDM tables and fields.
It is indeed that we have 2 results (so not an order and a result) from 2 locations at the same time. So for example if you take a swap from a wound, you can take a swap from the lateral edge and from the medial edge. From with the results can both be negative for certain bacteria.
@MPhilofsky in your option 1 you suggest the Specimen table, but I do not see a connection between the results in de measurement table and a row in the specimen table, so how could you combine the loction with the result then?
Well, with OMOP CDM v5.4, you would have to use the Fact Relationship table to join two facts.
However, additional fields might be added to the Specimen table in v5.5 which would enable linking it to the Measurement table. @Christian_Reich, is adding *_event_field_concept_id and *_event_id to the Specimen table on the roadmap for v5.5?
No use case. Look: unless you treat a single patient (which we don’t in OHDSI) or if you want to compare medial and lateral swaps and their diagnostic value (extremely unlikely), it is irrelevant. What’s relevant and might be used in a cohort is the bacterial infection. Whether you have one or two or ten such records of the same day makes no difference.
Bottom line: Just write the records, skip the gymnastics of trying to tell the exact story.