We have noticed an inconsistency in our review of nontraumatic vs. spontaneous injury. In some cases we make a distinction between them:
1296953008 |Nontraumatic rupture of extensor tendon of left hand (disorder)|
321371000119100 |Spontaneous rupture of extensor tendon of left hand (disorder)|
whereas in other cases we treat them as synonymous:
image.png
In determining the meaning of nontraumatic vs. spontaneous, we have found that nontraumatic means injury not caused by an external force and could occur with or without premonitory signs or symptoms, whereas "spontaneous" injuries usually refer to acute injuries without premonitory signs or symptoms.
The primary question is whether in general clinical practice this distinction has value, and if so, 1) should SNOMED explicitly separate out concepts where spontaneous injury is expressed as a synonym for nontraumatic and 2) should SNOMED create nontraumatic injury concepts as parents to spontaneous injuries where a nontraumatic parent does not currently exist?
17 Comments
Jim Case
From Monique van Berkum
Perhaps this has already been decided? I haven’t thought about it enough to know whether I agree, but the Editorial guidance below seems to imply that “Spontaneous injury” is distinct from |Non-traumatic injury|. 789750003 |Spontaneous event (event)| was added to SNOMED in 2020 and is used with DUE To 112 times in the stated view. Not all uses are for concepts that descend from |Non-traumatic injury|.
Even if these terms are used in clinical practice, I don’t think that most clinicians would appreciate the distinction between non-traumatic and spontaneous based the latter not having premonitory signs or symptoms. The guidance below implies that “Spontaneous injury” means injury with no underlying mechanism. If this distinction is maintained for injuries going forward, the difference between “Spontaneous injury” and |Non-traumatic injury| would need to be more clearly articulated.
Jim Case
The original application of this was to address issues with structural inconsistencies in the injury hierarchy in the QI project and while we did address the difference in the ed guide superficially, based on the FSN, I agree that the guidelines are not clear as to why the distinction is being made. Addition of the specific characteristics of spontaneous, vs. nontraumatic might make this clearer. From a clinical perspective, I think nontraumatic would most often be used, as it subsumes spontaneous injury, but then spontaneous injury concepts would be available when the distinction is desired. Not trying to bias the discussion regarding the direction of travel, just providing my opinion.
Jim Case
From Feikje Hielkema-Raadsveld
I have had a look through our translations and translation guidelines. For the most part, we have translated these terms literally, copying the English inconsistency and occasionally introducing some more inconsistency ourselves - as one does.
Spontaneous appears to be a negative concept, that you use when there is no external cause. In non-injury concepts, this excludes for instance infectious causes; in injury concepts, it is less clear what such causes might be. If there are no other identifiable causes for an injury other than a trauma the distinction between spontaneous and non-traumatic is moot and in those cases the terms are used interchangeably.
The Netherlands has an explicit guideline that says to translate spontaneous haemorrhage or non-traumatic haemorrhage both as PTN niet-traumatische hemorragie SYN spontane bloeding. I.e. the guideline specifically recommends non-traumatic haemorrhage and sponteneous haemorrhage as synonymous.
Non-traumatic rupture has usually been translated in the same manner: PTN niet-traumatisch, SYN spontaan. However, when searching through SNOMED's concepts concerning spontaneous rupture, I believe I saw some exceptions in cases where an iatrogenic event is another likely cause:
Prolonged spontaneous rupture of membranes - The other usual cause for ruptured membranes is iatrogenic and deliberate, I would hope a traumatic cause is quite rare; to distinguish between non-traumatic and traumatic would be quite confusing with respect to ruptured membranes.
Spontaneous rupture of esophagus (Boerhaave syndrom) https://www.ncbi.nlm.nih.gov/books/NBK430808/ Boerhaave syndrome is usually not truly spontaneous, but this term helps distinguish it from iatrogenic perforation. Vomiting is the most common cause, but any activity that increases intraesophageal pressure can result in this syndrome.
An injury with iatrogenic cause is certainly not spontaneous. But I am not sure: does iatrogenic count as traumatic, or as non-traumatic? Or neither?
All this I'm afraid is not particularly helpful in resolving on a course of action. I think there are many concepts, e.g. the haemorrhages, where spontaneous injury comes down to the same thing as non-traumatic injury and thus would make a proper synonym. But those are exceptions to the more general rule that a spontaneous disorder is more specific than a non-traumatic disorder; I fear we would have to identify the exceptions by examining them case by case. On the bright side, only 41 descendants of < 1119219007 |Nontraumatic injury (disorder)| have 'spontaneous' in a description. To answer Jim's original questions:
1) should SNOMED explicitly separate out concepts where spontaneous injury is expressed as a synonym for nontraumatic
Only if there is a usecase for that spontaneous injury as a separate concept. Please leave those haemorrhages as they are.
and 2) should SNOMED create nontraumatic injury concepts as parents to spontaneous injuries where a nontraumatic parent does not currently exist?
If it is a concept where spontaneous is in effect synonymous with non-traumatic, you would be better off changing the FSN or replacing it. Concepts such as 1088181000119100 |Spontaneous rupture of tympanic membrane of left ear co-occurrent and due to recurrent acute suppurative otitis media (disorder)| do refer to the more specific spontaneous and non-traumatic would I think just confuse. I have not managed to find a concept where an additional non-traumatic variant would be helpful.
Jim Case
Feikje,
Thanks for the comments. From what you have said, in many cases, the distinction between nontraumatic and spontaneous does not provide much added benefit. ICD-11 classifies iatrogenic injuries as traumatic (e.g. Iatrogenic pneumothorax is a subtype of Traumatic pneumothorax) as they refer to external causes. It is interesting that in the case of 1088181000119100 |Spontaneous rupture of tympanic membrane of left ear co-occurrent and due to recurrent acute suppurative otitis media (disorder), this concept does not comply with our understanding of spontaneous referring to a lack of premonitory signs. It is mapped to an ICD-10 complex map that references the residual class H72.9 Perforation of tympanic membrane, unspecified (with an index term of "Rupture, ruptured eardrum, nontraumatic). Nor is it modeled with a DUE TO = Spontaneous event. So there is an implied synonymy between nontraumatic and spontaneous in this concept.
It is still not clear whether the distinction provides benefit, but additional questions remain:
If they can be considered synonymous, what should the FSN and PT be? "Spontaneous" is used as commonly in clinical parlance as "Nontraumatic".
Does SNOMED CT "Need" to be consistent in this regard (i.e. terming)?
Should separate concepts that distinguish spontaneous and nontraumatic for the same condition be combined as a single concept?
Jim Case
From Jeff Pierson
Great discussion. Attached is an analysis, which may or may not be helpful.
Jim, I agree with Monique that having clear definitions should help us to determine if there is synonymy and how to move forward.
Also, all things being equal, I think “nontraumatic” is better as an FSN because “spontaneous” is harder to pin down.
Monica Harry
Jim Case - has there been a final decision on this yet?
Jim Case
No resolution yet.
Feikje Hielkema-Raadsveld
Jim Case Is this something we could/should discuss again at the next EAG? Or has a different follow-up action been planned?
Jim Case
I have not followed up on this. If I have time to review prior to the meeting in Korea I will add it to the agenda.
Jim Case
In looking again at the subtle differences between "nontraumatic" and "spontaneous", the consistent distinction that is mentioned is that spontaneous injuries occur unexpectedly, without any external forces, and without any premonitory signs (this was mentioned in the first comments of this thread), while nontraumatic injuries occur without external forces but may or may not have premonitory signs. Thus "nontraumatic" injuries are a supertype of spontaneous injury.
The clinical usefulness of this distinction is related to any further evaluation of a patient after the injury occurs. In spontaneous injuries, this may result in additional testing to try and identify any unknown underlying cause for the event. In nontraumatic injuries with premonitory signs, preventive recommendations can be made to address future injuries of the same type.
If the consensus is that this distinction is useful enough to maintain, nontraumatic and spontaneous descriptions should not appear on the same concept. It also suggests that for injuries, where a spontaneous concept exists, a nontraumatic parent concept should also be present, but not vice versa. This, however, does not hold true in non-injury concepts that are spontaneous, (e.g. miscarriage).
There are 199 concepts that use the DUE TO = Spontaneous event relationship, of which only 75 are subtypes of nontraumatic injury. Spreadsheet showing the 301 subtypes of 1119219007 |Nontraumatic injury (disorder)| , which also shows which concepts have both nontraumatic and spontaneous descriptions in included (download to review). A review of these concepts raises a number of questions regarding what qualifies a disorder to be assigned the primitive parent 1119219007 |Nontraumatic injury (disorder)|.
Ed Cheetham
Can I please put in a vote for the view there is no reproducible conceptual distinction between 'spontaneous' and 'non-traumatic'...
As others have suggested, aside from its physiological/'normal way of things' uses ('spontaneous respiration', 'spontaneous rupture of membranes' etc.) 'spontaneous' is more often used in the context of abnormal states and is one of several ways to say 'we don't know why', along with (but not limited to):
The diagram shared here indicates situations in current SCT data where these word pairs are used synonymously, also indicating where other words such as 'pathological' and 'non-traumatic' appear synonymously too. If the diagram is further launched in a browser (using the 'Raw' button) then clicking on each edge can give access to an example concept in the SI browser where the word pairs are used synonymously (I'm not sure about the 'cryptogenic--non-traumatic' pair but I think the others are valid).
The intention is to illustrate that these words, in particular 'spontaneous', 'cryptogenic' and 'idiopathic' are used quite interchangeably to indicate lack, or absence of, established causal knowledge for given phenomena, often intended to contrast with more common or frequently-encountered human-scale or 'understood' microscopic/molecular causes. It would seem wrong to attempt to attribute any more precision to these words in specific contexts. Oddly 'unknown cause' and 'unexplained' are in a little island of their own.
There are occasionally efforts to identify and highlight differences (e.g. here for a discussion on idiopathic vs. cryptogenic pulmonary fibrosis - published as synonyms in SCT) but for the most part the meanings seem to blur together as some variant of 'none of the currently known causes'. In the case of 'injury content the overwhelming likely cause is 'traumatic' so 'none of the known causes' is treated as synonymous with 'non-traumatic'.
The impression I get is that any editorial stance (on these words/phrases) will be arbitrary - my preference would be that you leave released content unchanged unless there is good justification. Arguably in the 'spontaneous' vs. 'non-traumatic' situation 'non-traumatic' could be seen as better for FSN-naming (see question below about what is meant by 'spontaneous' in SCT) in that it at least commits to some explicit exclusion. There seems to be similar arbitrariness in ICD 11: ruptures of muscles, tendons and cysts tend to be 'spontaneous' (but defined as non-traumatic) whereas haemorrhages and fractures are named as nontraumatic so we are not alone.
Finally a question: What does SNOMED mean by its notion of 'spontaneous event'? If it means 'we don't know why' then it's hard to see what knowledge value, if any, the 'due to=spontaneous event' modelling adds, and if it doesn't mean this, what does it mean? Specifically, what is SNOMED's model of causation - does it allow for truly 'spontaneous' un-caused causes?
Thanks - Ed
Jim Case
Ed,
Thanks for your comments. The intent of this endeavor is to determine how to resolve the inconsistencies related to descriptions assigned as synonymous in some concepts but distinct in others. If there is no essential difference between non-traumatic and spontaneous, then we should not have separate concepts, although we do. I acknowledge that these terms are used rather loosely and I am assuming that because there is no definitive way to make the distinction, concepts were added based on the requested FSN as opposed to any objective difference between a non-traumatic or spontaneous event. From a modeling perspective, the value in adding the DUE TO "Spontaneous event" allows for these concepts to become sufficiently defined and distinguish them from the parent. For example, 17369002 |Miscarriage (disorder)| is modeled as DUE TO a spontaneous event to distinguish it from other forms of 363681007 |Pregnancy with abortive outcome (disorder)|. With regards to the meaning of "Spontaneous event", as noted above it represents injuries that occur unexpectedly, without any external forces. The underlying cause may be known although the event may not be "expected". For example, a rupture of an aneurysm may be a spontaneous event, even though the predisposition to such an event is known.
What I am hearing you say is that any attempt to make a reproducible distinction is fraught with frustration. This leaves us with some editorial decision-making and some resultant inactivations such as the apparent duplication demonstrated by the concept below:
1296953008 |Nontraumatic rupture of extensor tendon of left hand (disorder)|
321371000119100 |Spontaneous rupture of extensor tendon of left hand (disorder)|
Remaining questions include, a) when should non-traumatic or spontaneous be used for FSNs and b) should all non-traumatic concepts have a spontaneous synonym and vice versa? The primary issue would be to remove duplicates (if there is consensus that these terms are synonymous).
Ed Cheetham
Thanks Jim.
In response to your 'remaining questions', for (a) I would repeat my suggestion above that "...'non-traumatic' could be seen as better for FSN-naming...in that it at least commits to some explicit exclusion...", and for (b) I would say no - deal with it where it already occurs or is requested but don't propagate it over all the 'non-traumatic' data; there are dozens of other situations where word and phrase equivalence is incompletely manifested in the English language reference data, and if people need it for improving search sensitivity then this should be done by search-specific means.
In other areas where 'spontaneous' appears as a synonym, I'd suggest that (given the troublesome polysemy of 'spontaneous') you look at an approach where, if there is an equally acceptable alternative word then that word should appear in the FSN. Where 'spontaneous' is the only clinically adopted word you probably have no choice but to use it.
In the handful of cases where distinct concepts currently exist (presumably as fall-out from a recent wave of concept additions) the 'spontaneous'/'nontraumatic' pairs can be treated as duplicates and justifiably same_as merged (preserving the older conceptId as active).
Regarding the 'spontaneous event' definition and the due_to=spontaneous event modelling stuff: to my mind this just doesn't hold up to scrutiny. You seem to be saying that the distinguishing characteristics of 'spontaneous events' are that they are (a) 'unexpected' and (b) that the 'cause may be known' but is definitely not an 'external force'. OK (leaving aside the likely fact that 'no external force' is a judgement of scale - I'm sure health and weakened vessels and membranes are bursting all the time due to external forces, it's just most of the time those forces are really small). The 'unexpected' aspect introduces notions of 'risk' and 'chance' of occurrence, and might arguably be invoked for any illness for which there is not a demonstrable macroscopic ('human scale') external cause (or chain of causes back to some arbitrary point). I appreciate that this thread is motivated by 'non-traumatic' notions, but current due_to=spontaneous event modelling already goes well beyond this. Plenty of illnesses present 'out of the blue' ('unexpectedly') - and either immediately or over time their 'cause may be known'; why would they not also be modelled with due_to=spontaneous event? I will continue to argue the due_to=spontaneous event modelling should be removed - it adds nothing to the knowledge contained in SNOMED CT, and the fact that it allows a number of concepts to be sufficiently defined may, in fact, be more of a risk than a benefit.
Matt Cordell
Before trying to find a distinction between non-traumatic / spontaneous, I think it's useful having a look at what the distinction between traumatic / non-traumatic attempting..
Within SNOMED 417746004|Traumatic injury (disorder)| is defined simply as "Disease due to traumatic event".
417163006|Traumatic or non-traumatic injury (disorder)| just has a GCI "Disease due to traumatic event" OR "Disease with associated morphology damage".
For |Traumatic injury (disorder)| - I assume not including a morphology (or causative agent) might support non-physical injuries. e.g. 47505003|PTSD|, but that's not a subtype.
The intersection of Traumatic injury and "Mental condition" (<417746004 AND <384821006) all appear to be head injuries.
The Proximal Primitive Modelling approach, also requires every concept to have "Due to Traumatic event" stated. Even though it doesn't really add any value. Only 24% of concepts are defined with something more specific than "Traumatic event"
And when there is something more specific for the event it's often not saying anything more than what the injury. Requiring a precoordinated event to do so...
The events are often only traumatic because injury. Without the injury, the patient was just fishing, measuring out some acid, or having a picnic.
When I wake with a 360450007|Strain of neck muscle (disorder)| - was sleeping a traumatic event?
So the definition of traumatic injury should be reviewed. It's probably better defined with some GCIs (based on morphologies and causative agents, and sometimes event). Are burns, bites, and lacerations ever NOT traumatic injuries?
Surely 219423008|War injury due to fireball effects of nuclear weapon| is a traumatic injury?
Adding "traumatic morphology" (that subsumes such morphologies) would work too.
So...
It seems like |Nontraumatic injury| is really asserting "a disease of some morphologic abnormality*" that was not caused by an external agent or due to some event.
* Note, 63% (19/30) of the "non-traumatic morphologies" are not even types of <37782003|Damage|
((<1119219007.116676008) MINUS <<37782003);and so only classed as types of |Traumatic or non-traumatic injury (disorder)| because they've been stated as subtypes of |Nontraumatic injury (disorder)|.My feeling is Non-traumatic/spontaneous/pathlogical are generally saying the same thing (no known external cause suspected) and the different adjectives might depend on the condition, but generally interchangeable.
I would also expect every Non-traumatic disease to have a traumatic counterpart.
I think the whole area needs a little work.
Jim Case
Agree that the whole area needs another review; however the traumatic injury hierarchy was only recently revised after extensive discussion with the Event-condition-Episode group over many months as well as extensive internal discussion.
From the point of view of this specific discussion, what I am hearing from most of the comments is that the ability to make a reproducible distinction between spontaneous and non-traumatic injuries is doomed to failure. The reason they were created in the first place was based on the requested terms from members, so they were being used somewhere. A summary of what can be gleaned from this discussion as a pragmatic resolution (not solution) for the issue (open to additional discussion)
Feikje Hielkema-Raadsveld
I think that's a sensible solution Jim.
Matt Cordell
I think that seems like a reasonable approach.
re: 5 - Review the need for "due to spontaneous event".
This just seems wrong... The "spontaneous injury" IS the (spontaneous) event. If it's actually due to some event wouldn't it be traumatic?
Just looking at the current 200 concepts defined with "due to spontaneous event" raises questions...
Half these appear to be from the "KP donation"
I don't think it's neccessary to revert concepts to primitive if this is property is dropped. They could be stated as subtypes of 1119219007|Nontraumatic injury (disorder)| instead.