Events, Conditions, Episodes Project Group meeting minutes 05-09-2016
Date
2016-05-09
Time: 19:00 UTC
GoToMeeting Details
https://global.gotomeeting.com/join/175028173
Recordings of previous meetings
https://kponline.webex.com/kponline/ldr.php?RCID=f44964120d4c561a71b52955df82adef
https://drive.google.com/file/d/0B2PpoaCzRXDJRWxOMENjdFlFZk0/view?usp=drive_web
https://drive.google.com/open?id=0B3L9Aniw9oukOWFmSFVlN3NKbjA
Attendees
Yongsheng Gao
Former user (Deleted)
Penni Hernandez (Unlicensed)
Apologies
Former user (Deleted)
Objectives
See agenda below.
Note: meeting today not recorded
Discussion items
Item | Description | Owner | Notes | Action |
|---|---|---|---|---|
1 | Welcome and role call | BGO |
|
|
2 | Review of previous meeting minutes and actions | BGO |
|
|
3 | Revisions to editorial guidelines | BGO
|
| Action
|
4 | Secondary disorders | YGA |
art6301_Sequela (finding) and sequela of disorders-elaboration phase v1.1 | Action Modeling of secondary disorders discussed A proposal was discussed for modeling secondary disorders as disorder X associated with disease (disorder). Secondary disorders are generally used as groupers for disorder complications modeled using more specific associations such as due to, after and caused by and thus associated with would be permissible in this context. This approach however will fail to autoclassify disorders secondary to material objects, procedures, events, etc. Currently, disorders related to material agents are manually asserted under (primitive) secondary disorders. Disorders related to other disorders are modeled and fully defined as disorder X associated with disease. Possible approaches:
|
5 | Plan for development of domain-specific templates for modeling combined disorders | ALL |
| Action Subject areas to start with (in order)
Additional areas to be based on identification of high frequency domains from KP and NHS GP/FP refset Next meeting to begin developing guidelines for dental complications |
6 | Drug overdose | YAO | YAO discussed modeling of drug overdose Drug overdose may be represented in SNOMED CT in 3 ways:
| Action Recommendation made to move all unspecified drug overdoses to event hierarchy and rename specified overdose disorders as Injury due to X overdose (disorder). Further discussion required re. overdose finding as an abnormal (elevation) of a drug on a laboratory test may not be due to an overdose but rather to a metabolic abnormality. YAO to post Powerpoint file he presented today |
7 | Next meeting |
| The next meeting will be held on Monday June 6th at 1900 UTC |
|
Meeting Files
Previous Meetings
Comments
David, perhaps this will help:
When looking at combined life phases, each individual life phase is temporally related to the other by one of three mechanisms:
1. Complete overlapa. These are the co-occurrent patterns during which one of the life phases is entirely contained within the course of the other.b. There are 8 co-occurrent patterns corresponding to Allen’s patterns, x starts y, x finishes y, x during y and x equals y and their inverses i.e y starts x…etc.c. Modeling each of these patterns by adding additional roles would be difficult and of limited use as it clinically would be difficult to differentiate among them.d. Therefore these patterns are lumped together into one (co-occurrent) and the important aspect is to capture the co-occurrence of both life phases by assigning them as parentse. When a causal relationship exists (x= cause, y=effect) x starts y and x equals y are the only co-occurrent patterns that are possible.
2. Partial overlapa. Allen’s pattern overlaps with – currently not represented in SNOMED CT
3. No overlapa. After – means never co-occurrent. b. Potential of creating new attribute to capture causality (due to and after discussed)c. Potential of approving before attribute requires discussion
4. x causes y is a special case applicable to a disorder + material agent only
Additional attributes representing temporal sequences such as during and during and/or after being considered for modeling procedure complications.
Bruce
Bruce, I had responded in email but then it occurred to me that I probably should have posted my thoughts and questions here. In any event, I found your explanations to be quite helpful.
1. I think I missed the point that co-occurrent means that one of the life phases is entirely contained within the course of the other, so I appreciate this clarification.
1.d. I think the point I’m having hardest time understanding is that co-occurrence can be adequately captured by making each separate life phase a parent of the combined disorder concept. Is this a consequence of the reinterpretation of the Clinical finding hierarchy as clinical life phases, which are occurrents (ontologically speaking)? Does an IS-A relationship between two clinical life phases necessarily imply co-occurrence? If so, does this hold throughout the hierarchy? In other words, is a clinical life phase concept always considered to be co-occurrent with all of its parents and ancestors? If a super-concept/sub-concept (IS-A) relationship necessarily implies co-occurrence, then I can see why you don’t need a separate CO-OCCURRENT WITH role in the model. So the phrase “co-occurrent with” only appears in the FSN, and there is no need to model it with a distinct attribute. Is that right?
2. So SNOMED CT currently plans to represent either complete temporal overlap or no temporal overlap at all. Is that right?
3.a. I can see that AFTER means never co-occurrent. Does this mean that X AFTER Y would be a sub-concept of X but not a sub-concept of Y, because making X AFTER Y a sub-concept of both X and Y would imply co-occurrence?
3.b. Are the choices for representing “due to and after” either a new attribute or a role group containing both AFTER and DUE TO? If so, has there been a decision that one approach is better than another?
4. I understand that “caused by” should only appear in the FSN when a disorder has a CAUSATIVE AGENT that is a material entity. I believe that the proposed editorial documentation suggested that a temporal relationship does not need to be represented in these cases. Is that right?
Perhaps we can talk a little more about all this on the next call.
Many thanks.
After looking at the proposed changes to the editorial guide for combined disorders, it occurred to me that listing the different possibilities in the following format might make things a little easier to understand. This format suggests that one think about a known temporal relationship before thinking about a possible causal relationship.
X BEFORE Y
(X finishes before Y starts)
X CONCURRENT WITH Y
(There is at least some temporal overlap between X and Y)
X CONCURRENT WITH AND DUE TO Y
X DURING Y
(X starts after Y starts and X finishes before Y finishes)
X DURING AND DUE TO Y
X DURING AND AFTER Y
(X starts after Y starts and continues after Y finishes)
X DURING AND AFTER AND DUE TO Y
X AFTER Y
(Y starts and finishes before X starts)
X AFTER AND DUE TO Y
X CAUSATIVE AGENT Y
(Y is a material entity)
(The means of exposure or introduction of Y is not significant)
(No temporal relationship is specified)
My first question, of course, is whether this representation is correct. Assuming that it is correct, I’m still not entirely clear on how these cases would actually be modeled. One approach is to treat the bolded phases as distinct roles and sub-roles. I suppose an alternative would be to put a temporal relationship attribute and the existing DUE TO attribute inside a role group. first and a possible causal relationship second.