SNOMED CT Editorial Advisory Group

Page tree

Inactivation ReasonAssociationDefinition

Classification Derived Concept 

REPLACED_BY

PARTIALLY_EQUIVALENT_TO

The inactivation reason “Classification Concept” explicitly states that the inactivated concept originates from the “closed world” classification paradigm and as such is inappropriate content for use within the clinical record. The inactivated concept (A) may be linked to one or more active concepts (B and/or C…) using historical associations as determined by the type of classification concept and associated guidance given below.

Please note that this inactivation reason replaces "Limited Component" and historical association "WAS_A".

Guidance
  • Identification of "Closed World content"
    The donor 'terminologies' to SNOMED CT were often themselves an early evolution of classifications of mortality and morbidity used for statistical analysis and reporting. To support the closed world view it was necessary to include options that catered for disorders that were known but not currently included within the classification. These concepts can usually be identified by the inclusion of “other specified”,  “not otherwise specified”, or “not elsewhere classified”, etc. While many of these concepts have already been inactivated, a few remain, and others such as "...without.." and concepts that are complex combinations of clinical phenomena still exist and will over time be removed.
  • Editorial guidance for specific flavors of classification concepts
    As each flavor of classification concept is identified, editorial guidance will be provided giving the reasons for inactivation and advice on how to assign an appropriated "REPLACED_BY" historical association.
    Each type of classification concept will then be identified within the authoring tool by a drop-down list e.g.:
    • The concept contains "Not Otherwise Specified"
    • The concept contains "Not Elsewhere Classified"
    • The concept contains "Without"
    • The concept contains an "and" or "with" representing a clinical statement composition (conjunction)
Not Otherwise Specified
  • Concepts that contain the ending "Not Otherwise Specified" are used within the closed world environment to enable the recording of conditions that exist but are not currently specifically identified within the current version of the classification in use. In the open world view adopted by SNOMED CT, this concept could represent any disorder represented by the concept without the addition of the "Not Otherwise Specified".
  • The recommended REPLACED_BY concept for "xxx not otherwise specified" is "xxx" as this represents our current open-world understanding of the meaning of the inactivated concept:
    • "xxx" may already exist as a current supertype or be found elsewhere within SNOMED CT
    • If "xxx" does not exist, it should be created as a new concept
The concept contains an "and" or "with" representing a clinical statement composition
  • This is where the “AND” (or “WITH”) represents clinical statement composition - a record of when independent clinical phenomena are co-occurrent in a particular patient. Many, but not all are derived from classifications, and for the purposes of patient care, it is recommended that each disorder is recorded individually. 
  • Represent each of the individual elements of the conjunction by separate clinical concepts using an association type of “PARTIALLY_EQUIVALENT_TO”. This signifies to the implementer and end-user that the single concept MUST be represented within the clinical notes by ALL of the PARTIALLY_ EQUIVALENT_TO targets.
  • Annotation
    Authors may wish to add additional notes for specific inactivations.

Examples

Concept with "Not otherwise specified"

Classification concept:

Resolution

The concept contains an "and" or "with" representing a clinical statement composition (true conjunction)

Classification concept:

Resolution


Resolving sequences of Historical Associations

The intention is that functionality to resolve sequences of Historical Associations will normally be seamlessly integrated into the tooling so as to present to the user the appropriate updated historically association to be allocated.

Whenever an already stated “REPLACED_BY” target itself also becomes inactive - whether at the same release or later, identifying the replacement for the original concept, should follow the combinatorial logic stated below.

Combinatorial Logic - REPLACED_BY

(A) REPLACED_BY (B) and (B) SAME_AS (C) implies (A) REPLACED_BY (C)

(A) REPLACED_BY (B) and (B) REPLACED_BY (C) implies (A) REPLACED_BY (C)

(AIntEd) REPLACED_BY (BIntEd) and (BIntEd) MOVED_TO (CNRC) implies (AIntEd) MOVED_TO (CNRC)

(A) REPLACED_BY (B) and (B) POSSIBLY_EQUIVALENT_TO (C, D) implies (A) POSSIBLY_EQUIVALENT_TO (C, D)

(A) REPLACED_BY (B) and (B) WAS_A (C AND D) implies (A) WAS_A (C AND D)

Note: Once MOVED_TO the NRC we (SNOMED International) have no knowledge of what has happened to CIntEd

  • No labels

14 Comments

  1. Matt Cordell

    The guidance recommends that EQUIVALENT_TO associations be between

    194733006 |Mitral and aortic stenosis (disorder)| and  
    79619009 |Mitral valve stenosis (disorder)|
     60573004 |Aortic valve stenosis (disorder)|

    This is wrong. These three concepts are not equivalent.

    A referenced component should not have more than one active “EQUIVALENT_TO  " at any given time. If it's equivalent to two things, those two things are surely equivalent too?
    There's also have consequences for "data healing" activities on longitudinal data stores so the cardinality of these associations is important.

    Also, I get that there are some classification concepts that use “and” when they possibly mean “or”, but I’m not convinced this is an example of. Is the intention to get rid of all the conjunction concepts? That seems overzealous as many are very valid, and used. For example, do you prevent people from recording “Diarrhea and vomiting”. It’s easy to say Information models should handle this, but that’s not a simple implementation exercise without causing a poor user experience.

    Groupers like “Menopausal and postmenopausal disorders (disorder)" I agree are not appropriate. But I don’t think a blanket rejection of “co-occurrence/conjunction” is the right move.

    Similarly for “without”. Although we can’t (currently) model negation, Negative assertions are just as important as conjunctions. “Placenta praevia without haemorrhage" “Refractory migraine without aura".

    We obviously don’t want to precoordinate every combination but many are useful, and removing existing concepts is not without downstream impact.

  2. Paul Amos

    Hi Matt Cordell,

    A number of issues here, I'll try and respond to each in turn.

    ICD has made our lives more difficult by occasionally using "and" when they mean "and/or". In this specific instance, we have interpreted 194733006 |Mitral and aortic stenosis (disorder)| to mean true conjunction i.e. the patient has both mitral AND aortic stenosis. I believe that this is how the average clinician would interpret this concept if they came across it in the patient's notes.

    Based on this interpretation, if we were to inactivate 194733006 |Mitral and aortic stenosis (disorder)| it would be clinically unsafe for us not to provide BOTH 79619009 |Mitral valve stenosis (disorder)| and  60573004 |Aortic valve stenosis (disorder)| as necessary replacements and insist that BOTH are represented within the patient's active record.

    If the interpretation of the "and" was that it actually represented "and/or" then 194733006 |Mitral and aortic stenosis (disorder)| would be considered to be ambiguous as we would not know whether it represented mitral or aortic or mitral and aortic stenosis. In this instance, we would use POSSIBLY_EQUIVALENT_TO for each possible interpretation and the user should choose ONE concept that represented the meaning at the time the record was made.

    We are not suggesting that all instances of a true "and" should be inactivated and a number of our editorial policies provide for exceptions for all the reasons that you have highlighted. However, as a general principle, we believe that there is a strong case for recording each disorder that a patient suffers separately as each will likely require separate and different clinical management.

    If you feel that “EQUIVALENT_TO" does not adequately portray the meaning we have ascribed to it, we would be happy to consider suggestions for an alternative word/phrase.

  3. Nicola Ingram

    Hi Paul Amos 

    Hope you don't mind a classification comment:

    • The concept contains "Not elsewhere Specified"

     Just noting that "Not elsewhere Specified" is not in the ICD classification or SNOMED CT (active or inactive).

    Prior to their inactivation the ICD-10 '.8' codes existed as 'other specified' concepts in SNOMED CT (majority inactivated as part of bulk Limited inactivations (WAS_A)_ when  .8 and .9 NOS codes were both inactivated in the NHS circa 2005 but not separately identified as 'other specified' which I think this is now proposing in the form of a drop-down list).  There are no 'other specified' concepts remaining in the current SNOMED CT but 3331 inactive 'other specified' concepts that may be impacted by inactivation of their active historical association targets.

    "Not elsewhere classified"  checked the ICD-10 instruction manual it says that: The words ‘not elsewhere classified’, when used in a three-character category title, serve as a warning that certain specified variants of the listed conditions may appear in other parts of the classification. For example:
    J16 Pneumonia due to other infectious organisms, not elsewhere classified.

    Also NEC is used in the ICD-10 index to direct the user if they cannot locate that particular condition, usually pointing to the .8 or 'other specified' codes to use as the residual, or bucket codes when the condition cannot be classified elsewhere. None exist in the current SNOMED but about 400 inactive 'elsewhere classified' concepts.

    • Not Otherwise Specified 

    Whereas the .8 or 'other specified' codes match your description of "enabling the recording of conditions that exist but are not currently specifically identified within the current version of the classification in use" (see my comments regarding NEC above) the NOS codes are the .9 codes in ICD. The ICD-10 instruction manual has this explanation: "The letters NOS are an abbreviation for ‘not otherwise specified’, implying ‘unspecified’ or ‘unqualified’."

    Sorry for being a geeky clinical coder (smile)

    Nicki

  4. Paul Amos

    Hi Nicola Ingram,

    No apologies needed. I agree with your definitions and the typographical error has been corrected.

    The list given is just examples and not exhaustive so we may add some to cover OPCS 'bucket' codes etc. These may be used to support the few remaining active concepts and, where agreed, to update legacy historical associations.

  5. Nicola Ingram

    Thanks Paul Amos although I should have said NOS is also defined as the unspecified form of the disorder so the definition you have does not really fit the NOS. The 'other specified' or just 'other' can be defined as used within the closed world environment of ICD-10 to enable the recording of conditions that exist but are not currently specifically identified within the current version of the classification in use. 

    Example  (.8 and .9)

    A03 Shigellosis
    A03.0 Shigellosis due to Shigella dysenteriae
    Group A shigellosis [Shiga-Kruse dysentery]
    A03.1 Shigellosis due to Shigella flexneri
    Group B shigellosis
    A03.2 Shigellosis due to Shigella boydii
    Group C shigellosis
    A03.3 Shigellosis due to Shigella sonnei
    Group D shigellosis
    A03.8 Other shigellosis
    A03.9 Shigellosis, unspecified
    Bacillary dysentery NOS

    1. Jim Case

      Nicola Ingram,

      So as I understand this, NOS <> Other specified and Other specified means the condition exists, but is not in the existing list, but NOS just means the condition is not recorded to any more granular level of detail than the grouper parent? i.e. a .9 means either unspecified or NOS? 

  6. Nicola Ingram

    Jim CaseYes you are correct  - 'other specified' means the condition is qualified but cannot be captured elsewhere in the classification (usually the .8 code in a category but not always) whereas the unspecified or NOS 'not otherwise specified' is the basic disorder with no qualifiers (the .9 code the majority of the time) - the category grouper parent in the ICD-10 classification is not always a guide as it can sometimes be named in relation to the other ICD categories.

    Category S09 Other and unspecified injuries of head

    S09.9 Unspecified injury of head

    Injury of:

    • face NOS
    • ear NOS
    • nose NOS


    NI note: Use of NOS - all the information available is, for example,  Injury of face and no further detail regarding the site on the face or type of face injury - assign S09.9.

  7. Nicola Ingram

    Jim Caseso as not to confuse this reason with e.g. classification of the terminology using a classifier may I suggest this is renamed to 'Classification schema entity'?

     My own personal viewpoint having had some small part in the Limited WAS_A inactivation in c 2004-5 (checking the ICD-10 codes) as part of the NHS Classifications Service and with no NOS or 'other specified' or 'not elsewhere classified' terming remaining in current SNOMED may I question the value of starting now 16 years+ dividing these inactive 'limited' concepts (originating from .8 and .9 in ICD-10) into separate ICD categories (this adds complexity to SNOMED and we may not gain in added value in the data).

    Nicki


    1. Jim Case

      Nicola Ingram,

      I am not proposing that we make any distinction between the .8 and .9 categories. They are all classification concepts and do not belong in SNOMED.

      1. Nicola Ingram

        Jim CaseIt was this section above (if I am reading correctly) that I thought seems to be basically dividing some of the inactive content into the 'other specified' (.8) and 'unspecified' (.9)  (and other ICD code types e.g. not elsewhere classified ):

        "Each type of classification concept will then be identified within the authoring tool by a drop-down list e.g.:
        The concept contains "Not Otherwise Specified"
        The concept contains "Not Elsewhere Classified"
        The concept contains "Without"
        The concept contains an "and" or "with" representing a clinical statement composition (conjunction)"

        1. Jim Case

          I think we could handle the .8 and .9 concepts with a single explanation of "Residual class"

          1. Nicola Ingram

            Agree Jim Case Residual class or Residual category or Residual category class possibly.



          2. Paul Amos

            Nicola Ingramand Jim Case

            The list of "classification" types is open for discussion and the list provided just a set of examples that users can recognize. The list is likely to depend on whether or not we wish to revisit historical inactivations. I would agree that if we do not wish to make improvements/adjustments to specific classification categories such as NOS then there would be little point in including them in the active list.

            However, I do believe that any category included should be described specifically and accurately so that those using them have all the information necessary to implement them appropriately within their system.  

  8. Matt Cordell

    Thanks Paul Amos, I think you've already got the appropriate association in the statement:

    "it would be clinically unsafe for us not to provideBOTH79619009 |Mitral valve stenosis (disorder)| and  60573004 |Aortic valve stenosis (disorder)| as necessary replacements and insist thatBOTH are represented within the patient's active record"

    My suggestion would be to use the existing "replaced by". It seems more logical to say X is replaced by Z and Y, rather than X is equivalent to Y and equivalent to Z.