Stream: terminology
Topic: Codes for disease known not to be in specified set (SNOEMD)
Morten Ernebjerg (Jan 14 2022 at 08:26):
Suppose I am coding sub-types of a given disease X with SNOMED (e.g. diabetes, which has the sub-types type-1, type-2 etc.). Let's say there are 5 possible subtypes of X, viz. A, B, C, D, and E. However, for a particular purpose, I only want to explicitly capture the occurrence of A and B - for the rests, I only want to capture "patient has a different type of disease" (neither A nor B, so C, D, or E). The parent code X is of course also implied, but using that is too vague because it also includes A & B. I want to be able to code all cases uniformly as conditions, so I want the code for "this type of disease but not subtype A or B" to mean exactly that and not be a generic "other" code. In effect, I want a "union code" saying "either C or D or E". Is there some way of doing that (short of defining a custom code with that exact meaning)? E.g is there some of way of using a condition with the code set to X and then elsewhere using other codes to indicate (in a structured form) that the subtype is known not to be A or B?
Michael Lawley (Jan 14 2022 at 08:28):
Why not just allow people to say C D or E, then just treat them as a group for your use case?
Stopping someone from providing the extra detail (when known) is somewhat of an anti pattern.
Morten Ernebjerg (Jan 14 2022 at 08:30):
Agreed, but this is something that has already been fixed (pre-existing guidelines).
Michael Lawley (Jan 14 2022 at 08:34):
There is no way to do this with snomed as-is. The best you can do is define a new code - F - that is a child of X and the parent of C, D, and E.
Morten Ernebjerg (Jan 14 2022 at 08:38):
Can one design such a code (within SNOMED) "on-the-fly" or do you mean one should submit that as a suggestion for a new code to SNOMED?
Michael Lawley (Jan 14 2022 at 09:25):
It would have to be a new code for SNOMED. There's no good way to specify disjunction (or negation) via post coordination.
Given that this is a guidelines driven requirement, it's probably reasonable for the new code to be created.
If you absolutely must do this via post coordination then you could construct a Situation With Explicit Context code that says "X known present" and "A known absent" and "B known absent", but that's pretty horrid for data analysts to consume.
Morten Ernebjerg (Jan 14 2022 at 09:40):
OK, that sounds sensible - thanks @Michael Lawley !
Morten Ernebjerg (Jan 14 2022 at 10:53):
Thinking more about this: Is there a standard way, in SNOMED or ICD-10, of creating conjunctive codes? I.e. if I want to say "the patient is known to have either disease A or B" - again in cases where there is no parent term representing this particular set of codes.
Last updated: Apr 12 2022 at 19:14 UTC
