Page tree

Date: 2018-10-16

1600 - 2400 UTC

0900-1700 PDT

1200-2000 EDT

1300-2100 Argentina time

Zoom Meeting Details

SNOMED Int'l Editorial Advisory group  

Please join my meeting from your computer, tablet or smartphone:

Topic: SNOMED Editorial Advisory Group Face to Face Meeting

Time: Oct 16, 2018 1600 UTC

Join from PC, Mac, Linux, iOS or Android:


  • Obtain consensus on agenda items

Discussion items

1Call to order and role callJCA


Conflicts of interest

Notice of recording

JCAGRE - Contractor to SI, Principal in TermMed

3ECE UpdateBGO

EAG accepted proposal to eliminate the use of "Co-occurrent and Due to" pattern in favor of "Due to".

  • Bruce Goldberg to develop editorial guidance for co-occurrence (when to use).
  • Monica Harry to update Ed guide with new combined disorder pattern guidance.
4Allergy and Intolerance updateBGO

EAG members supported the proposed modeling.
  • Bruce Goldberg to implement Allergy model as proposed
  • Bruce Goldberg to update contact hypersensitivity disorder to use AFTER as opposed to "Due to"
5Secondary diseasesJCA

There are a large number of disorder concepts that refer to "Secondary" or "Secondary to". A query was sent to the WHO ICD MSAC (Medical and scientific advisory committee):

  • It is unclear from the [ICD-11] definitions associated with the various conditions that use the term "secondary" as to whether it refers to simple co-occurrence or some level of causality, or both, as most of the definitions define secondary disorders as "disorder secondary to X" which is not very helpful.
  • This is important from the SNOMED point of view as we are trying to separate out causality from simple temporal relationships. If by "secondary"; ICD actually means a "Complication X due to disease Y", then we would reconsider many of the terms we have that currently make a distinction between ""Complication" and "Secondary".
  • There is also a question concerning the phrase ‘associated with’. As it is ambiguous, it is desirable that this phrase be deprecated.

Implications on modeling and terming of existing content as part of the QI project need to be discussed and recommendations provided to the content team.

Related SNOMED trackers:

Modeling of secondary diseases

artf6264-Complications - sequelae - secondary

Concept model for secondary disorders

artf6302-Review of Secondary X (disorder) concepts versus X associated with another disease

artf221501-metastatic malignant neoplasm

WHO ICD-11 guidance

2.3.5 ‘Due to’ and ‘With’

‘Due to’ is the preferred term for categories where two conditions are mentioned and a causal sequence exists. Other terms, such as ‘caused by’ or ‘attributed to’ are allowable synonyms. The phrase ‘secondary to’ is equivalent and may also be included as a synonym. 'Associated with’ is the preferred term for categories where two conditions are mentioned and there is no causal sequence implied.

There was also some work done in the Reference guide by the Morbidity Reference Group in partnership with Quality and Safety.

Coding from health care practitioner documentation of “causal relationships”

Sometimes conditions that have a causal relationship are clearly documented by the health care practitioner using terms such as “due to”, “caused by”, or “arising from”. These connecting terms indicate the health care practitioner has made a causal link between, for example, condition A due to condition B.  However, sometimes conditions are documented with connecting terms that are ambiguous for the coder such as “with”, “after”, “in”, and “following”.  When ambiguous terms are documented and it is not clear whether the health care practitioner means a causal inference or not, the clinical coder should code each condition separately and not link in a cluster.

The clustering (postcoordination) is a particularly notable new feature in ICD11 that has permitted the introduction of powerful new clinical coding mechanisms for capturing clinical information in dimensions such as:

  • quality and safety coding for healthcare related injury and harms (see 3 part model described in section {YYY})
  • the addition of clinical detail using extension codes
  • the specification of diagnosis type and diagnosis timing using extension codes
  • the comprehensive description of late effects (sequelae) arising from prior conditions (see section {ZZZ})
  • the description of inter related stem code diagnoses where there is a clear causal relationship

    For more information on causal inference in the context of quality and safety, refer  to section {XXX (currently: Causation in the context of quality and safety)}

From October 2018 MSAC Minutes:

  • For entities where the aetiology is specifically known, and the combination of condition and aetiology exists as a fully precoordinated entity, “Due to” is the preferred term. Other terms, such as “caused by” or “attributed to” may be allowed synonyms. The phrase “secondary to” is equivalent and may also be included as a synonym. This proposed solution may only be used, however, when the causal link is clearly defined.
  • Note, the phrases “secondary” and “secondary to” are not logically equivalent and may not be used interchangeably.
  • In concepts where an association is known, yet not as a known or proven aetiological connection, and the condition and association combination exist as a fully precoordinated entity, “associated with” is the preferred term. The connections “with” and “in” are not allowed combinations.
  • When considering neoplastic entities, the allowed term is “metastatic”, not “secondary”.
  • There was agreement that ‘secondary’ implies causality, but there was not agreement as to what “associated with’ means.

RE: Neoplastic diseases: IARC sees secondary tumor or neoplasm as metastatic.

  • Bruce Goldberg to bring topic of secondary including modeling and FSN terming to ECE.
  • Jim Case to develop list of "Secondary X' terms for review by EAG.
6Historical association refsetJRO

Revisit the intended meaning and operational consequences of the nine subflavours of 900000000000522004|Historical association reference set (foundation metadata concept)|

  • ‘historical associations’ are to be declared between inactive concepts and their candidate active (and so taxonomically reportable) substitute(s).
  • Sections within the RefSets Practical and Terminology Services guide mention that such associations exist - though they’re rather silent on what you can or should do with them.
  • No obviously visible guidance (e.g. in the new Editorial Guide) on how and why authors should decide what makes a ‘good association’
  • Not a lot of understanding of the possible consequences of a bad or missing one.

7Sources of truthBGO

8Observables discussionDKADiscussion on what to do with existing, often ambiguously termed, observables.

9Followup on clinical statement model project groupJCA

Summary of discussion from F2F meeting

1. The distinction between Finding and Disease has been and is a cause of confusion for modelers and implementers.
2. The use of implied context for the Clinical findings/Disease hierarchy causes issues for implementers in that other context-types are located in a separate top-level hierarchy.
3. We are currently using the Clinical findings hierarchy as both “Clinical entities” and “Assertions”.
4. It is desirable to have a “pure” clinical entities hierarchy that can be used to populate assertions (clinical statements). Potential names proposed:
a. “Findables”
b. Phenomena
c. Clinical entity
5. A number of the attributes of the Clinical findings concept model are context-type relationships.
6. It was generally agreed that SNOMED should evolve to include a “context-less” set of defined clinical entities that would support the population of a more robust and comprehensive “clinical statement” model.
7. A review of the various extant (and useful) clinical statement models should be undertaken to inform the structure of a SNOMED CT clinical statement model.
8. The current Situation with explicit context model is viewed as a starting point for the development of the SNOMED Clinical statement model.
9. A clear statement regarding the removal of support for the “Soft context” for Clinical findings and Procedures must be communicated to the implementation community.
a. Removal references to soft context from the Editorial guide.
b. Recommended that clinical entities would not be used directly, but only as a component of a clinical statement.
10. Post-coordinated expressions have a number of issues related to construction, determination of equivalence and reusability that make them less appealing as a solution to context.
a. Most large EHR systems implementations do not support post-coordination.
11. The current Situation model simply provided a way to move concepts that were context-laden, out of the ostensibly context-free Clinical findings hierarchy.
12. Logical negation is out of scope.
a. Does not conform with non-binary representations of presence or absence.
13. Any solution should be developed in conjunction with information model developers.
14. We need to develop an incremental approach to this change as it may be viewed as to dramatic for some users.

Potential Actions

  • Write a project charter. Should outline what the end goal of the project is and what the perceived benefits and potential detriments there might be. Name?
  • Propose the creation of a formal project group (Clinical statement project?). The initial though is to create two types of groups, a small, formal work group and a larger project group. These would be modeled after the groups in the drugs project.
  • Write the Terms of Reference for the Project Work Group and the overall Project Group
  • Identify potential members. What is the proposed size of the group. The bigger the group, the more difficult it will be to get consensus. However, without adequate representation, the more chance we will have of getting pushback.
  • Develop a draft strategy and the critical path for addressing the issues that we identify

a. Identification of the specific issues.
b. Predict the potential impact of the terminology
c. Outline that potential issues that might impact users and implementers
d. Develop mitigating strategies for minimizing impact.

  • Notify the Community of practice about the project group and its objectives

a. Solicit feedback from the CoP. That will be our consultation process.
b. Change or revise the terms of reference as needed from input.

  • Begin environmental scan for clinical statement models that can be used as starting points for comparison. Candidates include:

a. HL7 Clinical Statement model: (
b. FHIR resources: (
c. CIMI?

10Future meetingsJCA


  1. I just wanted to add a comment regarding Item 8 (discussed at the end). We've produced a couple of refsets to (attempt to) exclude these groupers. (I think I've mentioned before).
    One for Clinical Findings and one for Procedures.

    The clincial findings one contains those that I think David mentioned, that just aggregate a bunch of findings "Finding of color of hair (finding)" etc.

    The procedure one, is perhaps more contentious and slightly different in that it does contain "defined" concepts, but we've said their so general to not be clinical useful e.g. "Ultrasonic guidance procedure (procedure)" Why? Where?

    Both rely heavily on String patterns. And not fun to maintain. They're not perfect, but we've had positive feedback on them from implementers.

  2. Matt Cordell ,

    Would you be willing to share your exclusion refset?

  3. Certainly Jim, I've uploaded both our Procedure and Clinical Finding sets in this spreadsheet.

    The general intent of these is to discourage users from using the codes in clinical records, some might be contentious (particularly the among procedures).

    But some of the findings, could also potentially be misunderstood by clinicians as "positive assertions". Particularly things like the response/reflex findings.

    E.g. "Finding of response to sound (finding)" should NOT be recorded anywhere. And just groups all the "options". E.g. "Responds to sound (finding)" or "No response to sound (finding)".

    There's evidence some users have also been known to (mis)use such concepts as "observables"...

  4. Thanks Matt Cordell,

    These concepts are generally groupers used to "organize" the terminology into arbitrary classifications.  Other than that function, do you see any use for having these in the terminology at all, given that we now have a robust query language in ECL?  Removing them would certainly flatten the hierarchies, but for the most part they have additional IS A parents to classify under.

    1. Agree, the only purpose I can really think of is vendors who don't support ECL and can only do "single concept subtype queries". But even then, I'm not sure how use these concepts are, and you can still emulate a most of the ECL with SQL for these concepts.

      Flatter hierarchy might not be as aesthetic, but better than the noise.