OHDSI Home | Forums | Wiki | Github

LOINC or SNOMED for admission and discharge date

Hi everyone,

We are interested in mapping surgery admission and discharge dates in two separate observations. From what I can see in Athena there are two good choices:
LOINC: 3040041 (Admission date), 3042411 (Discharge date)
SNOMED: 4161014 (Date of admission), 40485383 (Date of discharge)

I was wondering if there is a preferable community aprooach to this so that our decision can be more interoperable with other data holders.

Thank you in advance.

Hi @adts,

Yes, Admission and discharge dates define the start_date and end_date of a Visit Occurrence or Visit Detail record. See more information here.

For my ETL’s I have assumed that the LOINC or SNOMED values are equivalent since they are both Standard concepts. An analysis would have to assume that either concept might be used to identify an admission date. It is not the case here, but I have found in other situations that when I try to limit myself to either the SNOMed or LOINC vocabularies there inevitably are values that I need that only exists in one or the other vocabularies so that it is often necessary to use concepts from both vocabularies.

Thank you for your response. That would be an obvious choice but in our case we want to be able to use discharge date as an index date and unfortunately it is not possible to use visit_end_date as index date in ATLAS from what I know of. That is why we have been looking into mapping these into two different concepts in the Observation table.

Ah, I see. I’ll tag @Chris_Knoll since this is an Atlas issue. If Atlas doesn’t have this feature now, the team might be able to add it. Or Chris might have an idea on a workaround.

Back to your question, any standard concept_id will work to represent the data. The preferable approach from a community perspective is to still represent these Visits in the Visit Occurrence or Visit Detail table. No one and no studies will know to look for Visit data in the Observation table. Folks look for visit information in the Visit tables

1 Like
t