2016-04-19 Editorial AG Meeting Minutes

Updates on previous issues

JCA briefed the group on slashes in FSNs, use of URLs in definitions (see slide below).

JCA said DOIs are persistent even if URLs change, so the guidance would require use those instead of URLs. There would also be a note in the Editorial Guide on how to find them. ISBN was another unique, persistent standard for representation that the organization could use.

JCA said the intent was not to use them as the sole source of a definition, but rather as a reference so that people could go to the source of the definition. Currently, he said, there was a text definition with no source, so this provided a way to point to the source. 

JCA said he had a note to figure out what to do if the source was no longer available.

• Determine policy on what to do if the source of a DOI or ISBN definition is no longer available.

IGR said there needed to be some guidance written up on GMDN terms for the linkage table. He said there was a proposal underlying the concept model for devices that would assist with definitions of the content. There would still be some need for some text definitions, but that would be discussed with the Members. He confirmed that there was an action item to come up with some Editorial Guidance for naming GMDN devices.

TMO noted that slashes were prevalent in the product hierarchy too.

Review of drugs, products and substances

TMO showed a slide presentation on several topics:

Substances

TMO said approximately 100 concepts representing role had been retired from the Substance Hierarchy because they were showing erroneous relationships and presenting other problems. 

The activation of the new Authoring Tool, TMO said, meant that it was hard to retire concepts, so some of that had been postponed until the July 2016 Release.

LOINC had been doing some interesting QA of substances, so that was providing some clean-up work, TMO said. 

Drug Model

TMO said the RFPs were published on the web and had passed the close date. IGR said it was a 2-week turn around. 

Drug Content

TMO showed the slide below (and other slides that are not shown).

She said product strength involved a high level plan to move concepts into a new module. The team was also working with IHTSDO Technical Services to do some of the work in batches.

Changes in content would begin in development in May 2016. There would be intermittent previews for review and comment supported by the new authoring tool and terminology server. January 2017 would be the release of an updated hierarchy.

Devices

TMO said the team had been working on migration of the linkage table maintenance to the Mapping Tool. It was receiving and processing monthly updates from GMDN. It used to be twice a year, and that resulted in a significant lag in seeing the updates in a SNOMED CT Release. 

Domain Range Changes

TMO said currently did not use SCT as attribute values to define concepts in other hierarchies, and they looked into whether they could do that and why the decision had been made. It had been an intentional decision in 2009 because there was no use case, but the situation had changed. She gave an example of vaccines adverse reaction. She described two different scenarios on a way forward (see slide below).

JCA asked if Option 4 should be reconsidered. 

There was a discussion of using algorithms to handle the changes that would be needed. 

• JCA said the action would be to talk with Yongsheng Gao on his view on the impact of both of the laterality options on the anatomy model.

Overview of Event, Conditions and Episode Policy Recommendations

BGO made a presentation. He presented the work plan, then spoke about the during, during AND/OR after to associated with role hierarchy.

JMI asked if BGO had spoken to the Anesthesia SIG on that because they had been doing work on that.

• BGO to speak to Andrew Norton about Anesthesia SIG work concerning temporal model.

JCA said he disliked "and/or."

KCA said he had discomfort with "associated with." If someone had diabetes and HIV, was a symptom associated with one or the other?

JCA said he was in favor of more robust temporal descriptions which could have a temporal association but not necessarily a causal relationship. KCA said users could look at patterns by data mining but clinicians should not presume a causal relationship. JCA said "most likely due to" is what the causation indicated. 

• JCA suggested that BGO bring back to the ECE group that relaxing the view of causality was, at least in KCA's opinion, not a good thing. If you want to identify co-occurance of 2 disorders and not specifically identify causality, the relaxation might lead to over-interpretation of causality. 

JCA said that "caused by" presented a problem because sometimes medical knowledge showed that they are not caused by, but his hope (perhaps naive, he said) was that if you say something is due to, then medical knowledge had proven that causation. 

On ECE guidance, Eric said they should determine whether to use the Oxford comma. JCA said editorial guidance for FSNs was a grammatical standard to be applied consistently throughout the terminology.

There was a discussion about trademark names.

• Jim Case said there should be an action to look at acronyms as trademark names for things like assessment instruments in order to come up with a policy. 

JCA said acronyms could be included as an alternative description, but currently they would not be in the FSN, but that could change.

JCA asked if the ECE had definitive recommendations they wanted to see applied to the Editorial Guide? BGO said no, but the recommendations would be revised. JCA asked if, in the meantime while the recommendations were being revised, there was guidance in the Editorial Guide and a fair number of Co-occurrent  and Co-occurrent Due To that were being added, so what was the recommendation from the ECE group for the editors, because there was a backlog on those items. BGO said many more terms used Co-occurrent rather than Co-occurrent Due To because people were hesitant to assign causality, so he thought the approach might be to move more towards that and eliminate the use of Due To. He thought the guidance should be if one was not sure of direct causality to use Co-occurrent. JCA asked if the ECE would be making that into a formal recommendation to put into the Editorial Guidance? BGO indicated that it would. JCA encouraged them to prepare that guidance quickly so that it could be implemented.

• ECE to make formal recommendations for Editorial Guidance on Co-occurent and Co-occurrent and Due to.

JCA wondered aloud if the team would have to look through all the Co-occurrent and Co-occurrent and Due To's to make sure they were not over-modeling. JCA said an analysis should be done. BGO said he would do that.

• Bruce Goldberg to do an analysis of existing Co-occurrent and Co-occurrent and Due To to make sure they were not over-modeling.

BGO asked if they were okay with stating in the Editorial Guidance that Due To is strictly due to direct causality. JCA said that was the proposal that KCA was making. PAM asked if there was guidance on the level of "knowing" in order to use "due to." BGO said it was based on a literature review.

KCA proposed using Due To as an attribute that can be used in describing relationships in instances but not the term as a precoordinated expression in SNOMED. The connection is then made by the physician and not SNOMED. Leave it in the realm of instances and clinicians. JCA said there had been other contexts like this, and he asked if there was a class that represents the universe of instances of Due To. If you say yes, you can identify causality in an instance, should there be a class that represents that. Due To does not say 100 percent of time it's due to, just an instance. KCA agreed with that, but said that by putting the concept in, it's ripe for misuse. He raised the question of when is a finding a disease? He said it was a value judgment. We wanted physicians to be able to make that value judgment, but if it was enshrined in the terminology, it put the burden in the wrong place. KCA said ideally they would want to get rid of most or all of the Due To's and allow physicians who really needed to use Due To to use it but to not give them a shortcut that would affect the validity of the data. 

• Jim Case said there should be an evaluation of Due To's, and that would be one of their many things to follow-up on.

Break

Content Tracker Review

JCA said the Content Tracker had been moved from Collabnet to JIRA. He showed some graphics that showed status. 

Groupers

JCA raised the issue of what a grouper really meant and what they were allowed to do and not do. If there were grouper items that did not belong there, how could you remove them? What was the priority was this? Was the description and goals sufficient for that particular tracker item?

GRE spoke about his and thoughts on the matter.

• Guillermo Reynoso - JCA asked GRE to put his grouper analysis on the tracker item so it could be used to develop the final approach.

JCA agreed that they should be fully defined concepts with no children.

GRE agreed to be the owner of that one since he was already working on it.

Non-ASCII characters

(such as en-dash, em-dash...). KCA indicated that he thought the topography should move into the current age and not just keep old conventions. GRE commented as well.

Moving between hierarchies without retirement

KCA asked how heavy the proposed policy was, because there did not need to be a lot of non-sense work. JCA agreed and noted that the policy just had to be an explanation of why something could be moved. JCA said he thought it was just an editorial decision so needed to communicate to people about why it was happening and put it in the release notes. Further discussion followed.

Word order variants policy

JCA noted that the policy needed to address specific requests for word order variants that originate from external members. Was natural language specific enough for an FSN?

JCA raised the issue of whether SNOMED was an interface terminology or a reference terminology. If it was interface, then it was underspecified in the representation of descriptions.

KCA and JCA agreed that there was a perception out there that the core terminology was supposed to have both, but there in fact was a division of responsibilities, so providers could add their own sets of descriptions for interface, as Kaiser had done. JCA said people should be informed that whenever they needed a new term, it did not need to go into the core; instead they should be encouraged to use the underlying structure of SNOMED to support that. That eliminated the need for mapping because they would be SNOMED terms. JCA said that wouldn't be a bad guidance document, telling people that the opportunity existed for them to enhance SNOMED far beyond what IHTSDO could support by utilizing the structures underneath. 

• Draft guidance document to tell people that every new term they needed did not need to be in the core, but they could instead enhance SNOMED far beyond what IHTSDO could support by utilizing the structures underneath. 

There was a discussion about namespace identifiers. JCA noted that IHTSDO sometimes promoted extension materials to the core. KCA pointed out that some people had told him that they did not want to make patient-friendly extension if IHTSDO could then claim all the IP. He recommended finding a solution to that problem. 

• Look at the licensing wording to ensure that namespace identifier holders are not discouraged from developing extensions for fear of IHTSDO claiming the IP.

Modularization of SNOMED CT

KCA gave a presentation. He said modularization of SNOMED was underutilized. Currently there were only 2 modules: metadata module and everything else. Modularization was a frequently used principle in the design of software and hardware, and he explained some reasons why. Then he gave some examples of possible SNOMED modules:

KCA asked the opinion of the group if SNOMED should become more granular?

JCA said modularization of SNOMED would almost be a paradigm shift for the International Release.

Emma M said in the UK, you had to take the International Release, the UK clinical extension and the UK drug extension, so although there were advantages of modularization, it would be hard on implementers. You would have to take a lot of modules in order to get anything useful. 

PAM asked how interoperability would be managed. There was a discussion about that and how another level of complexity would be required.

KCA said he thought for SNOMED to be successful, people would have to build value on top of SNOMED. 

An observer said that modularization would show a maturation of the product and be responsive the market. Quality might increase in modularized sections. He said the 2 Campbells in Nebraska had a great use case for modularization involving cancer.

JCA said the idea of the topic was to get the idea out there, and the AG would probably follow up with the idea on future meetings. Everyone, he said, viewed it as a potential benefit, but the roadmap of how to accomplish it was not clear. 

Lymphadenopathy issues

JCA said the topic involved the clinical interpretation/usage of terms vs. way they were represented in the terminology itself, so it was presented as an example for discussion.

JCA said it needed a clinical review because concepts were inconsistently modeled and needed consistent representation. 

JCA said there needed to be a tracker because a substantial number of concepts would be affected.

• Add a tracker on issues surrounding the lymphadenopathy topic.

Additional items: Changing of organism names 

JCA said organism names changed fairly frequently. When a request for an organism name came in, there were three options: (1) one to one change between one genus to another genus, retire the old genus, create a new genus, and the old description gets added as a synonym. (2) keep current concept with the obsolete FSN add new description (new name) and make that the preferred term (the idea is that the organism did not change, just what we called the organism. disadvantage: also have to change the parent tot he new genus). (3) just change the FSN and keep the same identifier. 

KCA said keep the identifier, move the current FSN to being a synonym, and add a new FSN that would be the current recommended name. 

JCA already editorial policy of not eliminating historical name because people still used them so they needed them, but hadn't addressed this particular issue, 3 use cases. 

Jim Campbell agreed with KCA. GRE also agreed with keeping the concept ID. 

JCA summarized the Editorial AG advice: keep the concept identifier, retire the current FSN, assign a new FSN and a new preferred term that matches the current name and retain the existing description as acceptable. GRE suggested keeping the previous FSN as active and creating a new FSN, could have 2 active so long as one was preferred. JCA said that involved a question for tooling - it would break the tooling, so they would have to submit a change request to tooling group. 

• Jim Case Request input from technical services: can tooling support two FSNs as long as only one is preferred?

Break

Discussion of timing of future AG meetings

The group decided to keep the meetings monthly.

Create 3 bullet points