To obtain input on the approval for use of the unapproved attribute 129453003 |Deficiency of (attribute)| to support modeling of deficiency content.
| Date created | 28 Oct 2025 |
|---|---|
| Action | Review the content |
| Provide your feedback | |
| Circulate as necessary | |
| Status | Open |
| Disposition | Awaiting feedback |
| Feedback by | 30 November 2025 |
The links within the briefing note to ââŚa number of SNOMED general content tickets and content trackersâŚâ held on Jira do not work. Please correct them so reviewers have access to all the materials.
Hi @echeetham thank you very much for pointing this out.
These were created pre-migration, hence the old links. In order to make it more easily accessible to all, the referenced JIRA tickets are now provided in an Appendix in the document which has been updated in this topic.
Thanks again for the feedback and apologies for the inconvenience.
Hi @jcase,
I reviewed the briefing note and to answer the three questions:
1. Does the proposal as outlined impact content in your extension?
This change appears to have a minimal impact on the US Extension with only a minimal number of concepts impacted and requiring remodeling work.
2. Does the proposal as outlined impact current implementations?
I donât see the addition of this attribute as having a large impact on current implementations in the U.S..
3. Does the proposal as outlined not address important considerations in expanding and applying the MRCM?
The move of this attribute into active use creates a long-term inconsistency in the terminology that implementors and researchers will find cumbersome. If SNOMED International moves forward with this change, we are setting up a situation where disorders like Hemochromatosis, Hypercalcemia, Hypervitaminosis, and other disorders caused by an excess of a substance will need to be modeled differently than a deficiency of disorder. It may be prudent to either add an attribute to represent âexcess of â or refine the modeling approach to mimic the interperts / has interpretation attribute pair functionality to represent these disorders. This will make for easier ECL and SQL queries when ask questions like âShow me all clinical findings caused by a deficiency of â or âShow me all clinical findings caused by an excess of â. I am also unsure if modeling using a deficiency of (attribute) would mimic the classification results if we were to use a version of interprets / has interpretation to represent an excess of disorder scenario.
For implementors sake, I think we need to address both sides of this coin at the same time for consistency.
I am on the fence whether it makes better sense to have attributes for âDeficiency of (attribute)â and âExcess of (attribute)â with a direct link back to substance OR if using an interprets / has interpretation attribute pair works better.
If we go the attribute pair route, we are relying on the author to add the appropriate observable entity concept at time of authoring in order to ensure the concept is modeled as sufficiently defined. On the surface, this seems like more authoring work with more points of failure than just have two distinct attributes.
There are pros and cons for both options, so we already know which ever path we choose will be wrong at some point in the future.
You bring up a good point about the contrary property of âExcess ofâ. This does indeed lead to an asymmetry in the representation of deficiencies and excesses. It is interesting that SNOMED has the unapproved attribute âdeficiency ofâ but not the contrary âexcess ofâ. While the proposal for the approval of âdeficiency ofâ addresses the need for a representation of the physiological state of the subject, in the case of âexcessâ disorders, the current use of the INTERPRETS/HAS INTERPRETATION for both findings and disorders relates to the observation of measurement rather than the physiological state, and we have examples of this duality, such as:
166702002 |Serum calcium level above reference range (finding)| with subtype
66931009 |Hypercalcemia (disorder)|
both of which are modeled with the INTERPRETS/HAS INTERPRETATION group.
The challenge is that using another attribute to define deficiency and excess disorders will result in a lack of subsumption under the clinical finding, i.e. hypercalcemia (disorder) would no longer be a subtype of Serum calcium level above reference range (finding). This is probably more accurate ontologically, but it would break the long-standing representation in SNOMED of disorders as subtypes of the related findings.
While this type of supertype/subtype relationship is common in the âexcessâ concepts, it is not represented in the deficiency types of concepts, for example:
238115004 |Sodium deficiency (disorder)| is not a subtype of
1153567005 |Serum sodium below reference range (finding)|
It appears that for deficiency concepts, there has not been an expectation for the disorder concepts to be subtypes of the findings. So, while the approval of the âdeficiency ofâ attribute does not makes things worse, it does nothing the address this non-isomorphic representation between excesses and deficiencies, nor does it answer the question as to whether these disorder concepts are proper subtypes of measurement findings of the same substance.
Current content in this area is even more confused than that, Jim!
Whereas (as you indicate):
66931009 Hypercalcemia (disorder)
is-A 166702002 Serum calcium level above reference range (finding)
âŚin the case of three similarly termed pairs of (finding) concepts, the relationship is inverted:
1153567005 Serum sodium below reference range (finding)
is-A 89627008 Hyponatremia (finding)
1187035007 Blood viscosity below reference range (finding)
is-A 47872005 Hypoviscosity (finding)
165462005 Plasma viscosity below reference range (finding)
is-A 47872005 Hypoviscosity (finding)
Meanwhile, many other similar pairs of codes exist with no mutual taxonomic relationship at all, including:
119247004 Hypoalbuminemia (disorder)
1153477009 Serum albumin below reference range (finding)
10399008 Hypochloremia (disorder)
1187054007 Serum chloride below reference range (finding)
190855004 Hypomagnesemia (disorder)
1179460004 Blood magnesium below reference range (finding)
While weâre looking to untangle the historic conflation of biochemical deficiency states with the disease phenotypes that they cause, we should consider also taking the opportunity to untangle the conflations of enzyme deficiencies with the excess of substrate/precursor substance(s) that they lead to and which are often the most direct pathogenic cause of the observed disease phenotype. Also, though this is seen rather less commonly, the conflation between the upstream deficiency of an enzyme and the resulting downstream deficiencies of the product substance(s) they are supposed to create, or excesses of products from other induced biochemical pathways that have converted the substrate excess into the pathological excess of a different product.
For example, 23501004 Arginase deficiency (disorder) currently has a synonym of Hyperarginaemia, whilst 47719001 Hypervalinemia (disorder) has a synonym of Valine transaminase deficiency. In the specific case of 23501004 Arginase deficiency (disorder), whose current synonym effectively implies hyperarginaemia to be inseparably and always present, dietary intervention can normalise or nearly normalise arginine levels.
See also:
58256000 Dihydropteridine reductase deficiency (disorder) Hyperphenylalaninaemia, type IV
61071003 Proline dehydrogenase deficiency (disorder) Hyperprolinaemia, type I
64852002 Sarcosine dehydrogenase deficiency (disorder) Hypersarcosinaemia
720940008 Hepatic lipase deficiency (disorder) Hyperlipidaemia due to hepatic triglyceride lipase deficiency
80908008 Ornithine carbamoyltransferase deficiency (disorder) Hyperammonaemia, type II
190859005 Hypophosphatasia (disorder) Alkaline phosphatase deficiency
273700000 Hyper-beta-carnosinemia (disorder) Serum carnosinase deficiency
57119000 Hyperammonemia, type III (disorder) N-acetylglutamate synthetase deficiency
58558003 Hyperlysinemia (disorder) Saccharopine dehydrogenase deficiency
Of course, splitting such conflations apart immediately leads to the need that Jon already raised for a matching design pattern by which to model excesses.
Finally, if weâre looking to achieve a single coherent model covering the full spectrum of values for which âdeficiencyâ is at one end and âexcessâ at the otherâŚwhat about the midpoint on that scale, such as might be needed to represent the biochemical states such as âeuthyroidâ?
I am very glad to see this comment as the deconflation of enzyme deficiency with the disease it causes was one of the instigating reasons for this proposal after having identified it back in 2014 when the original deficiency consultant terminologist paper was written. We had gone back and forth on how to best repesent this, but the lack of a meaningful way to represent the deficiencies was always a roadblock.
With regards, to midpoint or normal ranges, the same situation applies, i.e. observational measurements and physiological state. Should we taxonomically decouple the measurement findings from the physiological state of the patient. A you identiifed, we are quite inconsistent currently in this, and a clear decision based on the expectations of users is needed. However, at time, given the current inconsistencies, it appears that the user community is ambivalent.
This reminded me of some work that Kent Spackman was doing back when the Observable entity model was first beign developed in which he proposed a new hierarchy called Observation result, which clearly separated out measurement from phenotype.
Can I suggest that before we go too far down separating the ââŚconflation of biochemical deficiency[/excess] states with the disease phenotypesâŚâ we consider its impact on the relationship between SNOMED CT content and the recent intake of Orphanet-derived diseases? Many of these have been named using inclusive and accretive conventions resulting in much of the synonymy which is presented as concerning above and in the briefing note. Attempts to satisfy, simultaneously, an Orphanet âlumperâ approach with a SNOMED âsplitterâ approach risks producing something that is less usable by the clinical and research communities than the data in its current state. Iâm not arguing that there isnât a challenge here, just that - down in the detail - there is unlikely ever to be a âconsistent and reproducibleâ solution.
Good suggestion Ed. We will certainly look to resolve this with Inserm.
I want to start with saying that I thought this brief was well written and easy to follow.
But as others have said - this only address half (or a third) of the associated content. And doing so introduces further inconsistencies. It seems reasonable to expect these three concept to adhere to some common pattern:
A complimentary attribute like âExcess ofâ might go some way. But Iâm not sure what a similar attribute for the âwithin rangeâ state is?
Just focusing on âdeficienciesâ might be a way to limit scope; but it doesnât actually improve the terminology from a holistic perspective.
Comments from @jrogers and @echeetham also highlight my long-term frustration with the distinction between âfindings and disordersâ in the terminology (though maybe itâs just me?). I can almost reconcile the distinction if findings were strictly âObservation resultâ (per Jimâs comment).
But it still feels like an arbitrary (and confusing) distinction for most concepts.
These seem like one and the same. You need to âmake the finding to declare the disorderâ and the âdisorder is a description for somebody with this findingâ. The disorder is defined by the finding. Of course, somebody can have Hypocalcemia and not have had a blood test finding - but would they be diagnosed without empirical evidence?
(also note that neither of those âcalcium deficienciesâ are subtypes of 70241007 |Nutritional deficiency disorder (disorder)| or any of the other subhierarchies mentioned in the brief. A shortcoming of the terminology, not the brief)
The two examples used to explain the distinction are:
38341003|Hypertensive disorder| vs 24184005|Blood pressure above reference range| - I agree these are different.
But Hypertension is really persistent/long term high blood pressure. However, the concepts are modelled differently, with no shared lineage, and no indication of clinical course.
The other is
Where itâs the exact same finding for both âconditionsâ. But one is intentional the other isnât. Is it still a problem/disorder if the patient doesnât want to procreate?
Iâve rambled enough, but it comes back to taking a holistic view and consider tackling the big underling problems we all know exist.
Sorry I forgot to actually answer the questions in the brief also:
This wonât impact our extension content. We only have two concepts with âdeficiencyâ in the term - both relate for functional issues.
You can use this ECL to check other editions..
<404684003 {{C moduleId != 900000000000207008 |SNOMED CT core module| }} {{term = âdeficiencyâ}}
(Sweden, UK, US, NL, Ireland - look to have some concepts that might be in scope)
The changes are unlikely to impact implementations. Most implementations donât rely on the concept definitions. The modelling is currently inconsistent/unreliable to do too much with - so can only improve.
The problem the proposal is trying to resolve is certainly worthy. But see comments above.
I am grateful for the thoughtful responses that have come from the review of this proposal. I accept that the proposal only addresses one side of the issue related to the quantitative levels of biomarkers. Beyond the current scope of the proposal to add an attribute that would allow definition of deficiency concepts, it appears that several related issues have been surfaced. These include the following plus a proposed resolution:
1. The recognition that while deficiencies are addressed, excesses are not. There should be a consistent representation of ranges of measurements the includes, above, within, and below reference range findings.
a. We stipulate that deficiencies, excesses, and normal ranges should be included for all existing measurement findings where it makes clinical sense. For example, it may not be appropriate to represent normal or overproduction of all specific enzymes.
2. Taxonomic inconsistency in that some deficiency disorders are not subtypes of Nutritional deficiency disorder (disorder)|.
a. This is representative of the inconsistent modeling that would hopefully be addressed by the modeling pattern applied to deficiency disorders.
3. The use of defining attributes for deficiency disorders will result in different modeling patterns between disorders and measurement findings. This would result in a loss of the potential supertype/subtype relationship between the finding and the disorder, which is currently inconsistently represented in the terminology.
a. This is a topic that may require additional discussion and consensus. The representation of a deficiency disorder as a subtype of a measurement finding may not be appropriate if it is accepted that a measurement finding represents a point in time objective measurement of a substance, whereas a deficiency disorder represents a more persistent physiological state of a patient that may involve assessment of multiple clinical inputs. In some cases, it is possible to assign a disorder without a specific objective measurement.
b. Thus, a single measurement finding does not represent the persistent clinical state of the patient (e.g. glucose above reference range <> diabetes). Laboratory measurements are evidence for a disorder, not the disorder. Given the ontological characteristic differences between a finding (i.e. observation result) and disorder, there should not be a proper supertype/subtype relationship between them. The finding contributes to the definition of the disorder.
4. The use of the proposed âdeficiency ofâ attribute would result in the ability to deconflate deficiency measures from the resulting disorder in concepts where the deficiency measure is a synonym.
a. This is viewed as the resolution of a long standing issue.
5. For concepts derived from outside terminologies where the conflation of deficiency measure and resulting disorder exists (e.g. Orphanet), a mechanism is needed to represent the distinction that maintains the intended meaning of the source term.
a. We have already encountered this in our creation of Orphanet concepts in the international release. Recognizing this issue, our pragmatic solution is to ensure that any description on an Orphanet concept that references the deficiency leading to the disorder would have the word âdisorderâ added to the FSN and PT. (e.g. 1367655003 |Phosphoribosylaminoimidazole carboxylase deficiency disorder (disorder)|)
These resolutions will be included in a revised BN that both increases the scope of the proposal and addresses the identified issues.