Call to order and roll call
Chair, Jim Case (JCA) welcomed the group. Attendance is recorded on the main meeting page here: 2016-09-28 Editorial Advisory Group Conference Call.
JCA told the group that he would be changing the order of the agenda in order to finish some administrative items first.
5. EAG Self-Evaluation
The AG members agreed that the AG was functioning well, progress was being made, and the group should continue. There was general agreement about how more meetings were needed in order to make even more progress, but there was acknowledgement that it would be difficult due to busy schedules and time zones. GRE suggested holding 3 face-to-face meetings per year. JCA said it would be useful if more editorial changes being decided internally in IHTSDO could be brought to the Editorial AG so that AG members' comments could contribute to the discussions and decisions.
6. Goals for next quarter/year
KCA suggested the meeting in Wellington could kick off discussion on a longer term project of looking at the hierarchies for diseases, diagnoses, findings and observations and trying to find a strategy to reduce the confusion on what to use where. He said even getting a statement of the problem would be a great outcome for Wellington, and that could lead to discussion of options. JCA agreed that working on the structure of the problem would be a good first step, allowing them to then break it into smaller tasks.
PAM noted genomics was moving ahead quickly in the UK due to government financing, but it would be useful to have a more collaborative, international approach. JCA replied that genomics had been discussed internally but nothing had been progressed.
KCA said it would be very useful for staff to provide some use cases to the Editorial AG. JCA agreed that the team wanted to do that for future large-scale terminology development projects, making them strongly use-case based. He said typical requests for changes were very small, involving the submitter just needing a concept, rather than having a more robust, verifiable use case as part of the request.
KCA said that one of the problems with SNOMED was that it was supposed to be meeting the needs of lots of different groups, who were not using it in the same way. He noted that everything in Epic had to have one and only one code. It could not even handle present or absent. One idea, he said was that maybe SNOMED needed to be partitioned so that different use cases could be accommodated. Maybe SNOMED should not be one thing, it should be a collection of modules, so that for one use case you might need modules A, B and C, and for another use case you might need modules B, D and E.
JCA said that the AG members had provided some great topics to address in Wellington and also to work on in 2017.
3. Drug Model Content
JCA said in the discussions around the drug model, there had been substantial discussions around the use of universal restrictions, and the consensus had been that unless we were supporting the prescribing use case, then there was not a specific need to use the universal restrictions in the description logic at this time. That, however, had led to some issues involving drug products. He asked Toni Morrison (TMO) to describe the problem.
While TMO was pulling up the browser to provide some examples, KCA asked if the problem involved the ALL restriction. JCA said no, it was about the ONLY restriction involving a product that contains only that drug and not a potential mix of drugs. KCA and JCA discussed the differences in modeling when looking at combination drug products using the ALL versus the ONLY restrictions.
TMO showed an example. KCA asked about differences in interpretation of "product." Does "product" mean something packaged, or a manufacturable substance? There was a lot of ambiguity in that particular semantic tag. He noted that once something came off the manufacturer's assembly line as "product," it could not have a subtype. But the other use of "product" was for abstract things, like active ingredients. But both were labeled as "products."
GRE noted that ingredients were in the substance hierarchy. Abstract tablets dis not exist as things in themselves, but they were still useful in pharmacy models. He said the semantic tag could be clarified.
KCA agreed with GRE's points and said the AG could begin by addressing the Fully Specified Name issue, differentiating between an abstract product and a dispensable product. KCA then said that some "products" were abstract products because they could be combined, but then a dispensable product was also a "product," which could not be combined. He suggested setting a policy, and he said he could volunteer John Kilbourne to help with it since his team was already working on integrating RxNorm and SNOMED. Once the definitions were agreed to, then the data sources like RxNorm could be used to auto-generate from source material whether a product was abstract or dispensable.
KCA then spoke about capsules versus tablets within this context. He said the form could be the boundary between the abstract product, then the generic product, and then the trade product that is dispensed. The trade products, he said, might be left to the country extensions. The generic product, however, could and should be in the core. He suggested three layers of distinction: abstract product, generic product and trade product.
Emma Melhuish (EME) and GRE indicated that this was a good way of thinking about it, although the names might be changed.
KCA noted that the NHS and Australia had strong drug models, and IHTSDO could build on those rather than reinventing the wheel.
KCA said he did not support introducing the ALL restriction. EME said she just hoped for a consistent position because if IHTSDO's model were consistent, then NHS could link to it, but without consistency problems would arise. KCA agreed that consistency was important, adding that the ALL restriction had been debated for 5 years or more. JCA replied that the group appeared to have concluded that discussion. No one had been arguing that there had to be a universal restriction, he said, so he would develop some tasks, assign them, and bring them back to the group.
At KCA and GRE's suggestion, the group then looked at discussions recorded on this Confluence page: Proposed drug model. KCA verbally summarized the written discussion. He and JCA then said that Members should be informed about the issues because FSNs would change and they needed to figure out how to integrate their national extensions with regard to trade products. KCA suggested asking Member input on the naming of the FSNs to make sure the names were clear and unambiguous.
KCA then moved to approve GRE's recommendations a, b and c from Proposed drug model, though C would be amended not to ask NRCs yes or no (the AG had already decided yes), but to ask NRCs for their unambiguous form for naming. GRE's recommendations are reprinted below:
a) That it is made clear that existing “products” (e.g. aspirin) actual meaning is “product containing …”, in alignment with their modeling as sufficiently defined using existential restrictions and ingredients.
b) That the FSNs for those concepts are expanded to unambiguously represent that meaning.
c) That consultation is considered with NRCs as to whether this initial step would be acceptable. I understand it would likely be, as the pharmacist use case was not supported either way and therefore the implementations aligned with the need for universal restrictions were resolved by other means. Agreement to uniformly represent one of the meanings would facilitate the research and testing of alternatives to support the other.
GRE seconded the motion. JCA called for discussion on the motion. BGO asked a question about how it would relate to allergies. After some discussion about that, the AG unanimously agreed to the proposal.
4. Assessment Instrument Responses
JCA said the editorial policy currently allowed the addition of assessment instrument named, but did not allow responses to the assessment instruments.
KCA said there was already a way to represent the components of an instrument, but it got more difficult when someone wanted to copyright the instrument. He said LOINC had a process to make sure permission from the copyright owner was secured in order to include it in LOINC. He suggested that when someone asked for inclusion of a data element with an IP issue, staff at IHTSDO could work on obtaining copyright permission.
PAM said it would be nice if it were feasible to flag those as only to be used within the context of the assessment, not in the general record, independent of the assessment. He also said sometimes people were not tight in their descriptions, so the descriptions could be ambiguous or nonsensical. He suggested developing a manual to advice against making statements ambiguous. KCA said he supported that idea. He said to the extent that what was being requested met editorial principles, then he thought they should be added.
After further discussion, JCA said in this case they were creating instrument-specific value sets integrated into the core itself. They would be identified by these top-level groupers. That would open the door for a lot, he said. He summarized the conclusions: (1) for areas where there is no restriction on the IP, these probably should be added; (2) we need to develop an editorial policy, or an acceptance policy, for those things tied to IP-restricted instruments; and (3) we need to develop guidelines for the creation of responses on instruments that are in conformity with the URU policy of IHTSDO. The AG members agreed with that summary.
Conclusions and adjournment
GRE suggested that the items they had decided be noted as firm decisions so that they did not have to address them again. JCA agreed, adding that on the drug model Confluence space there should be a page that says the EAG policy decisions that have been accepted are... so that the decisions did not have to be revisited.
JCA thanked the participants and adjourned the meeting approximately 1 hour and 33 minutes after its start.