OHDSI Home | Forums | Wiki | Github

SNOMED Overhaul Discussion

Dear all,

I would like to invite you to a WG call dedicated to SNOMED Overhaul on October 31st. As the SNOMED Overhaul is in progress we would like to discuss some upcoming changes with the Community:

  1. According to multiple user requests (and because we’re missing the concepts like that in the proper Domain), we will revert the domain of Spec Disease Status concepts and bring them back to Meas Value (abnormal, normal) and Observation (malignant) domains.
    We would like to know whether Spec Disease Status domain is needed to populate disease_status_id in the Specimen CDM table? If so, we can create a separate vocabulary that will accommodate such concepts.

  2. We propose to make the descendants of Administrative procedure non-standard concepts in the Observation domain:
    In OMOP sense procedures are “activities or processes ordered by, or carried out by, a healthcare provider on the patient with a diagnostic or therapeutic purpose”.
    The above-mentioned concepts represent not procedures, but a number of administrative processes that take place during organization of healthcare:

  • patient admission procedures:
concept_id concept_code concept_name concept_class_id standard_concept_id domain_id vocabulary_id
4128582 305271006 Admission by dermatologist Procedure Standard Observation SNOMED
  • transfers:
concept_id concept_code concept_name concept_class_id standard_concept_id domain_id vocabulary_id
4294886 37729005 Patient transfer, in-hospital Procedure Standard Observation SNOMED
  • requests:
concept_id concept_code concept_name concept_class_id standard_concept_id domain_id vocabulary_id
45766931 704199009 Request for postmortem examination Procedure Standard Observation SNOMED
  • checking procedures:
concept_id concept_code concept_name concept_class_id standard_concept_id domain_id vocabulary_id
4307834 422759001 Pre-procedure evaluation check Procedure Standard Observation SNOMED
  • administration procedures:
concept_id concept_code concept_name concept_class_id standard_concept_id domain_id vocabulary_id
44807557 787581000000100 Medication monitoring administration Procedure Standard Observation SNOMED
  • QA procedures, etc (see full list in the attachment):
concept_id concept_code concept_name concept_class_id standard_concept_id domain_id vocabulary_id
4213600 80341003 Primary audit Procedure Standard Observation SNOMED

We would like to hear the pros and cons of this approach from the Community.

  1. We investigate concepts with procedure status = ‘Done’. These concepts belong to concept_class_id = ‘Context-dependent’ and have tag ‘situation’ in the source and belong to the hierarchy of Situation with explicit context. So, they do not carry the semantics of the procedure itself, but the status that this procedure has been performed, i.e. historical record of the procedure:
concept_id concept_code concept_name concept_class_id standard_concept_id domain_id vocabulary_id
4120075 287860003 Anastomosis of intestine revised Context-dependent Standard Observation SNOMED
37207730 151021000119106 Rabies vaccination given Context-dependent Standard Observation SNOMED
4124151 287637009 Dislocated wrist reduced Context-dependent Standard Observation SNOMED

These concepts were partially mapped to the ‘History of’ + procedure. We propose to de-standardize these concepts and map all of them to the ‘History of’ + associated procedure.

We would be happy to hear the Community’s opinion regarding the above-mentioned proposals.

Masha and the Vocabulary Team.

Related topics:

  1. About domain restrictions of MEASUREMENT.value_as_concept_id - #9 by andrey
  2. https://github.com/OHDSI/Vocabulary-v5.0/issues/741
    SNOMED administrative procedures.xlsx (18.7 KB)

TL;DR: if you care about SNOMED domains (#1), Family History (#7) or social context terms in SNOMED (#8), read this post, check out links and please-please come to the next Vocabulary WG meeting on Nov 14th.

We’ve been able to go through a lot of items on the call! We are coming back to discuss the remaining items on the next Vocab WG call November 14th noon EST.

A brief summary of what has been discussed:

  1. A systematic approach to domain assignment in SNOMED leading to changes in domains.
    Decision: gather more feedback
    Related post: Proposed changes in SNOMED domains

  2. Change domain for 3 concepts (abnormal, normal, malignant) from Spec Disease Status to Meas Value.
    Decision: have these terms in both domains unless conventions can be relaxed.
    Related post: Spec Disease Status vs Meas Value

  3. Deduplicate SNOMED drugs and dm+d (SNOMED UK) drugs by “erasing” the former.
    Decision: not to proceed

  4. Make administrative procedures such as “Admission by dermatologist” non-standard Observations.
    Decision: have them as Observations but keep Standard

  5. Change concepts like Rabies vaccination given to History of + value (to increase consistency of use of History of and to highlight the fact that these happened in the past)
    Decision: proceed as planned.

  6. Adding more mappings
    Decision: no objection here, we need more mappings :slight_smile:

What has not been discussed:

  1. Making concepts in classes a) Attribute, b) Linkage Assertion and Linkage Concept, c) Location (except for countries), d) Physical Force, e) Physical Object (except Devices), f) Social Context (except Relationship domain) non-standard.

Please check out the examples in the slides and come to the meeting if you care about these classes!

  1. Family history: creating a comprehensive hierarchy of family history.

Given that it is a chunk of work, we need to know if people use family history in their research at all :slight_smile:

1 Like

Hi everyone,

I’m not sure if this is the right place to ask such a question, but I’ll go for it anyway. We have around 200.000 entries that we want to convert to OMOP. It’s highly specific data about extracorporeal life support systems. In the data, you will find information about where the “majority of treatment” took place. From my understanding, we cannot map data anymore if you destandardize the location concepts. The only way this would work is as one big visit, though this might be inaccurate since we do not know about transfers to wards outside the intensive care units (where such therapy usually takes place). I see that the route Christian Reich described here (New Comprehensive Hierarchy for Providers, Visits (and Place of Service, Specialty, Care Site)) is fine for new data, where you know about the different transfers within your hospital premises, but it lacks a meaningful way to map aggregate data.

Any thoughts on that?
Thank you very much and have a great weekend!

The way to go would be to come to the Vocab WG meeting to talk about it. We got the message already and discussed it at one of the past WG meetings though so we will not de-standardize locations.