Content Model Changes

JCA explained that the were 2 domain and range revisions to discuss and 2 potential new attributes to discuss:

Domain and range revisions: (1) Clinical course - add additional disease phases, e.g. "in remission" and "latent"; (2) specimen substance - physical object (allowing for public and environmental health monitoring).

Potential new attributes: (1) During (e.g. "disorder X DURING procedure Y"), and (2) Has product role (needed to support the specific roles that are being removed as IS-A relationships from the product hierarchy.

"In remission"

On the addition of disease phases “in remission” and “latent”, JCA asked if anyone saw any potential problems. GRE said that several years ago he had done all the remodeling for the clinical course, and he could not see any potential impact. 

JCA said that he would anticipate them going in as siblings, so there would be no problem with subtype inheritance.

GRE said if it’s in remission, you are classifying it as a disease. JCA asked whether disease remission should be a finding or a disorder? GRE said it depended on whether one was considering it an episode of care or an episode of disease. If it is an episode of disease, in the future if it is out of remission it is still a disease. So, he said, it would depend on what one was modeling, but the ambiguity would be in the disease, not the remission. He agreed with JCA that they should be siblings.

KCA said he was not bothered about a disease in remission remaining in disease taxonomy. He added that the distinction between a finding of disease and disorder is unmaintainable so they should not worry about it.

PAM made a point about remission vs. cured, and some disorders like psychiatric would not use the term “cured.”

GRE made a point about the current situation of the patient being disorder but currently in remission, that is not a clinical course of the disorder as a characteristic of the disorder but an observation of the patient, so that probably needed to be distinguished. JCA said there was a tracker on the item and the exact wording would be needed.

Specimen subject as a physical object

On specimen subject being a physical object, JCA said in one of his previous jobs, he advised that any physical object could be a specimen substance, but some people did not like, for example, a tank being a specimen substance (or soil, water, air, etc).

KCA said he did not have a problem with a tank being a specimen substance.

JCA said the tracker was about extension of the definition of specimen subject, specifically for the benefit of public and environmental health monitoring.

GRE noted that device specimen is already a subset of specimen. JCA clarified that in many cases, it would not be a device. GRE suggested discussing it at the next conference call.

Potential new attributes to describe temporal context

JCA asked BGO to present his slides. Slide 1 about procedure complications:

“Combination disorder + procedure concepts”

 “Complications of procedures – time sequences: A complication (clinical finding or disorder) may not only occur after a procedure has completed but during the course of a procedure with varying degrees of overlap between the occurrence of the procedure and the complication. The time sequences between a procedure and its complication can be described using Allen’s interval algebra….”

BGO explained that Allen’s pattern included precedes, equals, meets, overlaps, contains, finished by, starts, and started by.

BGO went through additional slides, providing background on work that he and his colleagues had been doing. He said it came to a head when trying to define intraoperative complications using the associated with attribute.

GRE suggested creating another attribute for procedure period so that there would not be an overload on one attribute.

BGO said he and his colleagues immediately dismissed temporal context as not being useful and instead looked at occurrence, peri/pre/intra/post patterns.

Slide on Discussion of options:

• Temporal context – keep this for current purpose only
• Occurrence
  • Offers mechanism for representing peri/pre/intra/post patterns, Non-isomorphic with current “after” mechanism
  • Requires additional relationship for specified procedures
• Associate with extension
  • Non-isomorphic with current “occurrence” mechanism
  • Limited set needed (not all candidate values), however
  • Not clear which attribute names needed to handle necessary vagueness (e.g. does our “during” really mean the same as the Allen’s/EN12381 set?)

JCA noted that there were a number of requests for “before,” particularly with medications, like before meals and before bedtime.

Slide on Additional considerations:

• Discussion  re. attributes required
  • During
  • Before
  • Starts during
  • Contains, includes or bounded by (to represent “peri”)
• Test plan will need to be developed to investigate

GRE said they needed use cases that could be tested, particularly something that happens during the procedure. He said they needed to give a special meaning to “during” because part of the problem in SNOMED is that it considers an event to be a point in time, when it actually spans time. GRE suggested that BGO take the lead and that he (GRE) help with the technical aspects.

GRE then made a motion that they not add those additional values that overload the field and should instead get rid of those additional values from the range. JCA said that he would definitely agree that any addition of concepts that overload the meaning of the attribute should be removed. He suggested taking that as a motion.

KCA said he had a related comment, which was a concern about where the SNOMED model started and ended. There was pressure to try to push every patient visit into a single code. How should one draw the line and help implementers? He said he wanted to see a reconciliation between event, condition and observables and similar work being done in FHIR so he could see where the attributes are shared and map the two onto each other, which would mean fewer consequences of a single code representing something because it could be expanded out.

JCA agreed it was hard to find the boundary, but suggested that BGO take KCA’s suggestion of harmonizing the ECE model and Daniel Karlsson take a look at harmonizing the observables model with FHIR.

TMO gave a presentation. She said that both the substance hierarchy and the product hierarchy contained concepts for the role a product might play, and there was quite a lot of duplication. Some concepts representing the product role were being retired from the substance hierarchy, she said, like anti-glaucomal agent. Those concepts in the product hierarchy were associated by “is-a” relationships and were resulting in erroneous inferred relationships. She said the team needed to relocate those concepts. The current proposal, she said, involved the creation of a new sub hierarchy, probably within products (but the team would be looking to modeling group for advice on location), which would be primitive concepts representing the product role. They would also use a new “has” product role attribute that would use the new sub-hierarchy as a domain range. She asked if that seemed reasonable to the Editorial AG. AG members agreed that it did.

TMO went on: the attribute could be associated with a product concept at either the medicinal entity or medicinal form level, and concepts in the new sub-hierarchy might be used to fully define existing concepts in other hierarchies, such as procedure.

She said the team would need editorial guidance because “has” would not always fit into the “always and necessarily true” category. The topic, she said, would be discussed with the drug concept model group so the team could fully understand the use cases.

So, she emphasized, the team would be requesting a new attribute.

GRE, who had done a lot of testing of that sort of proposal in 2008-09, spoke for about 10 minutes on the topic. He indicated that he could help TMO find the work from 2008-09.