1 of 10

1

Date: 20th October 2025

Handling meaning changes in terminology

Yongsheng Gao

Modeling Advisory Group meeting

1

2 of 10

2

Background

    • Terminology meaning changes impact understanding and analytics
    • These changes are a common linguistic phenomenon
    • ‘Semiotic Triangle’ - Ogden-Richard
    • ‘Knowledge soup’ - John Sowa
      • The fluid, inconsistent, and loosely organized nature of human knowledge, which is constantly changing.
    • Understanding changes is crucial for effective communication
    • Addressing meaning evolution is important for clarity

3 of 10

3

Background

    • Concept inactivation reasons
      • Ambiguous concept
      • Classification derived concept
      • Duplicate concept
      • Erroneous concept
      • Meaning of concept unknown
      • Non-conformance to editorial policy
      • Outdated concept

    • Description inactivation reasons
      • Not Semantically Equivalent
      • Outdated component
      • Grammatical description error
      • Non-conformance to editorial policy

Historical associations in SNOMED CT

4 of 10

4

Potential challenges

    • Active concepts whose meaning in use may be more precise/grounded than the available terms can describe

    • Multiple definitions may have applied at different time of use

    • The most recent definition definitely doesn't even apply to all uses since the release of the most recent definition

Feedback from NHS:

5 of 10

5

Example - Sepsis

An international consensus definition for sepsis and related phenomena was published in 2016, and that document’s introduction points to earlier 1991 and 2001 consensus definition efforts. 91302008 has been in SCT since its first release, and almost certainly had a life in UK terminologies before this, so could well have been in use for the lifetime of all three definitions.

What can we know about each case/referent where 91302008 has been entered in the record?

Arguably ‘very little’; In the absence of metadata or related entries confirming the presence/absence of supporting criteria, the code and term can only invoke some imprecise thought of ‘sepsis’, which may or may not adequately stand for each actual referent (not to mention that 91302008 carries a synonym of ‘systemic infection’ which makes things even messier).

6 of 10

6

Example - End-stage renal disease

The first formal Chronic Kidney Disease (CKD) staging system was published in 2002 by the National Kidney Foundation, establishing the GFR-based classification of five stages. This framework was further endorsed with updates in 2004, 2012 and 2024. The CKD stages were added in SNOMED CT in 2008. 433146000 |Chronic kidney disease stage 5 (disorder)| is the final stage of CKD.

46177005 |End stage renal disease (disorder)| has existed in SNOMED CT since the first publication in 2002 as a subconcept of renal failure syndrome. The code for "End stage kidney disease" is thought to have originated from READ codes, specifically K05.., K050., and K0D... These codes, which were added between 1991 and 1997, referred to "Chronic renal failure", "End stage renal failure", or “End stage renal disease” and were derived from ICD-9 codes.

Consequently, older data predating 2002, and some post-2002 data, could lack the precise diagnostic criteria now used for CKD 5. While the two codes are likely clinically equivalent for new and future data, this equivalence does not extend to legacy data. These codes indicate the need for renal replacement therapies, e.g. dialysis or kidney transplant. The existence of two codes could cause potential discrepancies for analytics if one of them or both have been used.

7 of 10

7

Example - Pressure ulcer

1163215007 |Pressure injury (disorder)| has been created in SNOMED CT in 2022 based on the recommendations of the National Pressure Injury Advisory Panel (NPIAP) and adopted for the 2019 International Clinical Practice Guidelines on Prevention and Treatment of Pressure Ulcers/Injuries. The NPIAP nomenclature favors the use of pressure injury over pressure ulcer, due to confusion around the use of ulcer for two of the pressure ulcer stages which actually occur in intact skin.

New morphologies with text definitions have been created representing the various pressure injury stages. Existing concepts have been inactivated and replaced with corresponding Pressure injury concepts.

The SNOMED CT pressure injury disorder hierarchy follows the NPIAP terminology most closely but accommodates legacy classifications by including an ulcer morphology in the model as well as additional descriptions of pressure ulcer for pressure injury stages II - VI. The changes have been explained in the Editorial Guide.

8 of 10

8

Editorial Panel discussion

The meaning changes over time is an inherent challenge in medical terminology that can't be completely solved within the terminology itself

    • Clinical Evolution: This reflects ‘refinement’ of understanding rather than ‘complete’ meaning change
    • Temporal Context: The meaning depends on when the code was recorded, which should be captured through metadata and timestamps in EHR
    • Administrative Burden: Adding extensive annotation mechanisms would create significant overhead without clear user benefit
    • Implementation Reality: Clinicians don't typically view annotations or detailed metadata when selecting codes

9 of 10

9

Questions to Modeling AG

    • What is missing from the current inactivation mechanism in SNOMED CT?

    • What are the impacts if we do not address this beyond current mechanisms?

    • How should such changes be indicated practically?

    • How would this information be used for data entry and analytics?

10 of 10

THANK YOU

22