[Traumatic]/[Nontraumatic] Injury of [body structure] (disorder) - for review (note: same as [Traumatic] Injury of [body structure] (disorder) with 2nd optional role group added)role

Terming is difficult to standardize as many injuries as defined by a subtype of 37782003 |Damage (morphologic abnormality)|may not by named as nontraumatic injury of x body site

Peter G. Williams , please update template language.

Thank you.

Jim Case  is this one okay to go? Thanks, Cathy 

This  one is not quite ready to go.  The OCCURRENCE and CAUSATIVE AGENT relationships needs to be grouped with the morphology and finding site relationships to which they appply.

I made the suggested changes but I want to think about this some more. I included an optional morphology of damage to support the stroke model classifying under Nontraumatic brain injury without needing to use multiple due tos.

Bruce

I removed the first role group (containing damage (morphologic abnormality) ) as at this time it would only apply to nontraumatic brain injury. I will create a separate template for the latter.

Bruce Goldberg, currently the defining characteristic of 417163006 |Traumatic AND/OR non-traumatic injury (disorder)| is ASSOCIATED MORPHOLOGY = Damage.  So if we are talking about any type of injury, I think Damage is still relevant, whether stroke or not. 

 I thought we were going to do away with traumatic and/or nontraumatic?

If  X specified as traumatic, use due to nontraumatic event

If X specified as nontraumatic use due to spontaneous event

If X not specified use X

Fracture of bone - most currently are modeled as traumatic fractures \

Associated morphology fracture (after removing fracture morphology from under traumatic abnormality + due to traumatic event

Pathological fracture 

Associated morphology fracture (after removing fracture morphology from under traumatic abnormality + due to spontaneous event

Unspecified fracture of bone

Associated morphology fracture (after removing fracture morphology from under traumatic abnormality

Brain injury. again as this term is so frequently used in the literature, Imo it should be distinguished from brain injury and traumatic brain injury.

Bruce Goldberg ,

Very comprehensive and I think covers everything that I have come up against so far.  I am assuming that the proposed use of two DUE TO relationships is for CVAs where there is a spontaneous event DUE TO an occlusion/stenosis/obstruction?

I have been thinking a lot about fractures and since all of them have been classified as "Traumatic" for over 20 years, I am not sure of the benefit to users to begin to make a distinction where there can be both pathologic and traumatic fractures.  At the time of clinical presentation, one would be clear, the vast majority of the time, whether the patient suffered a traumatic event or not.  If the desire to create the "grouper" fracture is so that all fractures classify under a single hierarchy instead of sibling hierarchies, I would suggest that the added work is not worth the effort over educating users about how to find all fractures using two morphology classes in ECL.  I think it is something we can discuss at EAG for some more insight.  I will need to create a briefing note to outline the proposal, but want to make sure that you are in agreement (or I can browbeat you into agreement )

Hi Bruce Goldberg, what is the domain for this template? Is it 417163006|Traumatic AND/OR non-traumatic injury (disorder)|? Or are we going to inactivate all 'Traumatic AND/OR non-traumatic' concepts?  

The key issue is what does 'injury' means in SNOMED CT. Does it mean 'traumatic AND/OR non-traumatic' or 'traumatic'? 

There are over 2,300 concepts which contain the word 'injury' in their descriptions. Many of them are under 417746004|Traumatic injury (disorder)| but omitted the word 'traumatic' in their FSNs. It seems that we considered 'injury' means traumatic. However, it is hard to determine if it is traumatic or non-traumatic for those "injury of X body structure" concepts. Should they be interpreted as 'traumatic AND/OR non-traumatic' injury? Again, are we going to keep them or not? If we keep them, what would be the model of these concepts?

I think we can split this template to three templates for those 2,300 'injury' concepts and make them explicit in FSNs, e.g. Traumatic AND/OR non-traumatic injury of X, Traumatic injury of X, and non-traumatic injury of X.  The template for traumatic AND/OR non-traumatic injury could be modelled by GCI. The traumatic injury template can be modelled by any morphology with due to a traumatic event.  The non-traumatic injury can be modelled by the damage morphology due to a spontaneous event. 

Hi Yongsheng Gao.  This termplate supports, Traumatic, Nontraumatic and Traumatic  AND/OR nontraumatic injuries because due to traumatic event and due to Nontraumatic event are optional while associated morphology=damage is required. Thus, a Traumatic  AND/OR nontraumatic injury can be defined as isA disease associated morphology=damage without a due to event relationship and in that case I don't think a GCI is required nor is splitting this template into 3 separate templates necessary. The existing concepts with just "injury" in the FSN but modeled with as traumatic injuries should be renamed and the rest modeled as above. imo.

Bruce

Hi Yongsheng Gao. I think Jim Case  is under the same impression as I am. If you scroll up to his comment on 5/28:

"Bruce Goldberg, currently the defining characteristic of 417163006 |Traumatic AND/OR non-traumatic injury (disorder)| is ASSOCIATED MORPHOLOGY = Damage.  So if we are talking about any type of injury, I think Damage is still relevant, whether stroke or not. "

I think maybe we need a call to resolve this?

Bruce Goldberg and Yongsheng Gao ,

A call would be  helpful so we can more quickly get consensus on this.  I think we are close.

Hi all, Apologies for butting in. Speaking from a translation point of view: please don't change all 'injury' FSN's to traumatic injury! The inconsistencies between the FSN (Injury) and the definition (traumatic abnormality) has been a problem for years, and we have reported it but meanwhile translated according to the FSN, as recommended by Snomed International. I'm all for making these concepts consistent, but please do so by changing the definition instead... 

The morphology 'traumatic abnormality' is causing a lot of inconsistency at present, and using it does match this template. Is that morphology likely to be inactivated sometime soon?

Feikje Hielkema-Raadsveld, there is no intention of changing the existing  FSNs to include "Traumatic".  Injuries are being redefined using a DUE TO relationship as described above that allows for the inactivation of the Traumatic abnormality morphology, which is in progress and should be gone by the Jan 2021 International release.  

I do need to update this template based on discussions we have had since its inception e.g. due to traumatic event or spontaneous event shoud have an attribute cardinality of 1..1 and the required morphology of damage needs to be removed.

Hi Jim Caseand Bruce Goldberg It's great that you are improving this branch. Which 'due to' relation will a concept such as 125600009 |Injury of hip region (disorder)| receive? I noticed that 90584004 |Spinal cord injury (disorder)| has already received a 42752001 |due to| = 773760007 |traumatic event| ; but it's Dutch translation, 'dwarslaesie', can also be caused by a tumor (is that a spontaneous event?). And I'm fairly sure that holds true for the English term as well. 

I'd normally report this through CRS but it seems more pertinent here 

Feikje Hielkema-Raadsveld , we are defining Injury in line with the WHO definition which implicates an external cause of the damage to the body structure.  Thus for 125600009 |Injury of hip region (disorder)|, this will be defined using a 42752001 |due to| = 773760007 |traumatic event|  relationship.  Internal or physiological causes of damage will not be assigned a DUE TO relationship, but will be modeled as disorders with morphologies without a DUE TO Event relationship.  We are planing on using the DUE TO Spontaneous event modeling pattern for disorders that specify "Non-traumatic" or "Spontaneous" in the FSN or to differentiate disorders that are typically traumatic in nature, but due to underlying disease happen spontaneously, e.g. pathologic fractures, which are not caused by the underlying disease, but the  disease is a predisposing condition.

Hope this helps

Jim Case Jim, while I agree with your assesment, nontraumatic spinal cord injury (like nontraumatic brain injury) is a common notion in medicine and tumors are frequently cited as a cause. 

Grassner L, Marschallinger J, Dünser MW, et al. Nontraumatic spinal cord injury at the neurological intensive care unit: spectrum, causes of admission and predictors of mortality. Ther Adv Neurol Disord. 2016;9(2):85-94. doi:10.1177/1756285615621687.

Bruce Goldberg,  agree in principle, but I do not think we would be modeling tumors of the spinal cord (internal or external) as "Spontaneous events".  That type of clinical grouping seems to be a way to categorize things as intrinsic vs. extrinsic causes.  It would have substantial impacts if adopted universally.  We may want to look at the characterization of Non-traumatic disorders as those that do not have a DUE TO <<traumatic event as opposed to trying  to define them specifically as non-traumatic.  This is a complex area that does not lend itself to a clean solution as of yet.

Jim Case Modelling concepts called 'injury' as traumatic has a substantial impact on our translations, and we may not be the only country affected. At the very least prefer we will need all concepts with 'injury' in the FSN to be defined consistently. Is the idea that 'spinal cord injury' will be defined as traumatic, and will have a sister (not child) concept called 'non-traumatic spinal cord injury'? All 'traumatic injury' concepts to be renamed to 'injury', and all concepts that denote a 'traumatic and/or non-traumatic injury' inactivated?

Please don't underestimate the problems this is going to cause us...

Feikje Hielkema-Raadsveld , Nearly all injury concepts in SNOMED have been modeled as traumatic injuries since the beginning of SNOMED time.  This is due to the fact that all injury related morphologies were subtypes of 19130008 |Traumatic abnormality (morphologic abnormality)|, which was used to model the parent concept "Traumatic injury".  What we did not  have was the ability to model non-traumatic injuries properly, so things like "Pathological fracture" were also classified under traumatic injury.  So in this process we have not changed the original meaning of the injury concepts. 

We have tried as much as possible to adhere to the WHO definition of injury "...physical or physiological bodily harm resulting from interaction of the body with energy (mechanical, thermal, electrical, chemical or radiant, or due to extreme pressure) in an amount, or at a rate of transfer, that exceeds physical or physiological tolerance."

We are making a distinction where possible between non-traumatic (equating them with "Spontaneous") morphologies and traumatic ones, and creating new ones where needed, but our attempt was to keep the original meaning as modeled in the redesign.  Thus, for something like "Spinal cord injury", which was originally modeled as a traumatic abnormality, was maintained as such in the terminology.  We acknowledge the existence of the clinical use of "non-traumatic injury", but this is a negation that we have only recently begun to try and represent, and that in specific cases where the FSN specifically states it. The value of the newemodeling approach is that we can apply morphologies to concepts that result in physical damage, but are not DUE TO a Traumatic event, something we could not do in the past.

 I am sorry this causes you problems, but again, our attempt here was to retain the meaning of the  concepts as originally added to the terminology.  

Regards,

Jim

Hi Bruce, this topic came up today in the translation group discussion (see here: Trauma & injury). In French there is a significant problem to know how to translate "injury" versus "lesion" versus "wound". To know that all "injuries unspecified" = "traumatic injuries" for translation changes a lot of things. It's easier that way. We have been confused by the few concepts which are not cause by a trauma = accident but rather by external conditions and dared not translate injuries as traumatic even with the relationship to traumatic morphology because one is supposed to rely on the FSN as the source of truth and not depend on the relationships for meaning (even if in real practice, you often need to). I look forward to the final decisions you'll take here. As Feike said, this has not just an impact for you in modelling but in meaning for all of us who translate. This said, I think we need to develop a tight bond between translation group and modelling of templates. Ideally, we should publish the language specific templates hand in hand with the SNOMED international modelling/FSN/Syn templates if we are to make sure the meaning of the concepts is respected in all languages. 

I'm adding this example I just came across

33431000119109 |Lesion of oropharynx (disorder)|

child: 262664007 |Injury of oropharynx (disorder)|

child: 275489000 |Wound of oropharynx (disorder)|

The second is impossible to translate differently in French than the first unless one add "traumatic" to it. If not both are "lésion" in French, no other way.

I was wondering why I got so many misaligned concepts with this template and I see now that it only allows for one RG.  There are many traumatic injury concepts that require multiple RGs, (e.g., fracture-dislocations).  I am going to add the additional optional RG and ask the Yongsheng Gao redo the template language to reflect that addition.

Jim Case , I still think it might be preferable to reduce the complexity of this template by separating out traumatic injuries related to procedures.  The new template would be an exception for which 2 due to relationships are permitted if one wants to assert causality to the procedure. I can make the changes if you say yes.

Bruce

Created Injury of [body structure] [due to procedure](disorder) - new template for review

Yongsheng Gao ,

Can we update the template language to include  the second RG?

I think this revised template needs to be converted to 2 separate templates for traumatic and nontraumatic injuries. 

I have modified: [Traumatic] Injury of [body structure] (disorder) - for review and I have created: Nontraumatic Injury of [body structure] (disorder) - new template for review. Note that there is a duplicate template that I could not publish or now delete.

Bruce Goldberg, Yongsheng Gao,  

I am not sure what the status of this template is currently.  If we have a Traumatic injury template and a Nontraumatic injury template. We probably need a template for the supertype concepts that currently do not classify as traumatic or nontraumatic.  This template seems to exclude nontraumatic injuries and only has one RG.  Given that 417163006 |Traumatic AND/OR non-traumatic injury (disorder)| is not modeled with a GCI and is SD, how do we create a template for its direct subtypes?

Jim Case , I would delete this template as it has been superceded by [Traumatic] Injury of [body structure] (disorder) - for review. I thought we had decided that nontraumatic injury would be primitive and that Traumatic or nontraumatic injury would be modeled using GCIs. For concepts that can be either traumatic or nontraumatic would we not just assign Traumatic or nontraumatic injury as the parent?

Bruce

Jim Case With all the back and forth discussion we had I focused on Yong's suggestion for modeling Traumatic or nontraumatic injury using GCIs. If we are accepting the current model for Traumatic or nontraumatic injury (isA disease + associated morphology = damage) then the model for traumatic injury would need a 2nd optional role group containing associated morphology=damage for those traumatic injuries where the primary morphology is not a subtype of damage. I thought Yong's argument was that by modeling Traumatic or nontraumatic injury using GCIs, only 1 role group would be required. Please confirm my initial analysis and I will revise the Traumatic injury template.

Bruce

Looking at Traumatic or nontraumatic injury  in the TS browser or SCA,  am wondering why it is marked FD. I thought GCIs were never FD.

Yes, very interesting. If not FD, Traumatic or nontraumatic injury  does not subsume Nontraumatic injury. On the other hand if Traumatic or nontraumatic injury is modeled with 2 GCI axioms instead of 1, it will subsume Nontraumatic injury if left primitive. I am very interested in Yong's explanation.