OHDSI Home | Forums | Wiki | Github

How does the 'tender joint count' concept work?


In the context of EHDEN I am currently working on mapping some data to the proper vocabulary. Athena is being a bit help in this task, but I just stumbled upon something where I need some further help.
The data I am working on includes swollen and tender joint counts and SNOMED has matching concepts, see for example 376471000000106, but those concepts are ambiguous. My data always specifies the number of tender/swollen counts out of how many were actually counted. There are some standardised lists of joints, so there is no chaos. Doctors decide which of those they want to use and in research we make sure that we never compare counts that originate from the big list to those from the middle list, for example.

I expected to see those different lists implemented in the vocabulary, but they evidently are not. I am not sure how anybody can even use the current concepts without that information. You would be guaranteed to observe some anomalous effect where meticulous doctors have less healthy patients.

I considered hacking something together via the information about the maximum you can add to a entry, but at that point the people eventually using the data will have to be pretty careful to correctly interpret it, which kind of eliminates the purpose of this entire standardised mapping idea.

Am I missing something here? Currently I will probably just skip this information.

Exactly this! The OMOP CDM is analytics ready. No interpretation is needed. Standardization is the only way to perform network research.

I traced the ancestry of your SNOMED code/concept_id to parent and grandparent concept_ids to see if it had a related concept_id for joint count, but didn’t find anything. The Vocabulary team will know more.

I just actually discovered an even weirder property of this measurement now that I understand Athena better. The DAS28 is currently listed downstream from the two joint counts. I understand that they are linked since the counts are components of the DAS28, but I am surprised that this is the direction. It is also weird that the other components, like the C-reactive protein are not also linked.
Is it standard that the scores a measurement is used for are ‘subsumed’ by the measurement? This is pretty different to how it works for Conditions.