Hello all,
can we populate source_concept_id field with concept_id that have invalid_reason ‘D’ or ‘U’?
Such concepts in some cases even have mapping in concept_relationship table.
Thank you,
Irina
Nooooooo. Don’t. Post them here, so the Vocab team can fix things (or tell you it is already fixed).
Ok, thank you for quick response)
@Christian_Reich,
for example
C93.2 Subacute monocytic leukaemia, that’s deprecated in our vocabulary, and doesn’t exist in a current version of ICD10,
we have in ICD10 2008 year version
http://apps.who.int/classifications/icd10/browse/2008/en#/C93.2
so what we are going to do with these patients who had this “C93.2 Subacute monocytic leukaemia” before 2009?
@IYabbarova, there is a problem with deprecation here:
deprecation means two things unfortunately:
- nice concept that was deprecated because of reasons (people invent something new in a cancer diagnostics or similar) - but the thing is that somebody really had this condition
- something about
so, yes, the best way is to post them here and then we can make an analysis is there case 1 or case 2.
1 Like
Not sure I understand. These are deprecated Source Concepts. We map them to a fresh Standard Concept. Or where am I wrong here?
Yes, we do.
And initial question was
So according to my example:
can I put in a CDM source_concept_id 45755350 of those C93.2 code?
Oh. Thanks, @Dymshyts. You are right.
Yes, the Source Concepts can be deprecated, dead, burned, liquidated, annihilated, because they were valid at the time when they were recorded.
2 Likes
Can I clarify something here? I have noticed that, for example, some ICD-9 codes are marked deprecated, but they are were valid at some point and may be useful for research. I am trying to figure out what “deprecated” means in this context. Does that mean that the mapping is deprecated, and not the ICD-9 code? And there should be a non-deprecated version that reflects the updated mapping?
We treat deprecated concept differently from deprecated mappings. If a
concept is deprecated it means it is no longer valid, but at one time was.
So in doing an ETL if the code in the source data maps to a concept that is
deprecated we use the deprecated concept.
However, we view a deprecated mapping as an error in the mapping that was
probably never valid and therefore do not use a deprecated mapping. There
was talk about removing deprecated mappings from the vocabulary so that
they do not cause confusion, but I do not know what is currently being done.
“Deprecated”" sounds very dramatic, like somebody took an axe, cut it out and a lot of blood is splashing. All it means that as of today (SYSDATE) it is no longer valid. The consequences of that are:
- For Source Concepts it means - nothing. We still have to maintain mappings to Standard Concepts. You can safely ignore that.
- For Standard Concepts it means you can no longer use them in the data. All mappings have to be redirected to fresh ones. That could lag sometimes a little, and while it does you have no mapping (to 0). We are in the middle of building repeatable processes to really be on top of that and do it fast.
Makes sense?
Got it – thanks. So, for something like ICD-9 which is just a source concept, deprecated means nothing and I can ignore the deprecated status. I guess I am wondering what would cause it to be indicated as deprecated. But maybe not critical knowledge to have.
Don’t ask me. Ask the folks in the WHO and CMS who maintain these. ICD9CM is no longer maintained, so they are not touching it, but they are giving birth to new ICD10CM codes and they making others obsolete, so they can’t be used anymore.
OK. That comes from the vocabulary. I thought it was from OHDSI. That helps. We have all of the valid and non-valid dates for ICD9 so we can check it ourselves. Thanks.