Page tree

StatusIn progress



Termdescription typeLanguage/acceptabilityLanguage/acceptabilityCase significance
[Course] [Periods of life] [Process] Anemia due to [disease/finding ](disorder)FSNus:Pgb:Pci
[Course] [Periods of life] [Process] Anemia due to [disease/finding ]SYNus:Pgb:Pci

Concept model:

Definition status:  

900000000000074008 |Primitive|

Applies to

271737000 |Anemia (disorder)|

Template Language

64572001 |Disease (disorder)| : [[~0..1]]{[[~0..1]]  263502005 |Clinical course (attribute)|   = [[ +id (<   288524001 |Courses (qualifier value)| ) @course]]}, [[~0..1]]{[[~1..1]]  42752001 |Due to (attribute)|  = [[+id (<<  64572001 |Disease (disorder)|  OR <<  71388002 |Procedure (procedure)| ) @dueto ]]}, [[~1..1 @rolegroup]]{ [[~0..1]]   116676008 |Associated morphology (attribute)|   = [[+id(<<  49755003 |Morphologically abnormal structure (morphologic abnormality)|   ) @morphology ]], [[~0..1]]   363698007 |Finding site (attribute)|   = [[+id(<   123037004 |Body structure (body structure)|   )]], [[~0..1]]   246454002 |Occurrence (attribute)|   = [[ +id(<   282032007 |Periods of life (qualifier value)|   ) @occur]], [[~0..1]]   370135005 |Pathological process (attribute)|   = [[ +id ( <   308489006 |Pathological process (qualifier value)|   ) @proc]], [[~0..1]]   246075003 |Causative agent (attribute)|   = [[ +id (<   410607006 |Organism (organism)|   OR <   105590001 |Substance (substance)|   OR <   78621006 |Physical force (physical force)|  OR <   260787004 |Physical object (physical object)|   ) @causative]] }

GCI axiom 1:

Definition status:  

900000000000074008 |Primitive|

Template Language

64572001 |Disease (disorder)|  : [[~0..1]]{[[~0..1]]  263502005 |Clinical course (attribute)|   = [[ +id (<   288524001 |Courses (qualifier value)| ) @course]]}, [[~0..1]]{[[~1..1]]  42752001 |Due to (attribute)|  = [[+id (<<  64572001 |Disease (disorder)|  OR <<  71388002 |Procedure (procedure)| ) @dueto ]]}, [[~1..1 @rolegroup]]{ [[~0..1]]   116676008 |Associated morphology (attribute)|   = [[+id(<<  49755003 |Morphologically abnormal structure (morphologic abnormality)|   ) @morphology ]], [[~0..1]]   363698007 |Finding site (attribute)|   = [[+id(<   123037004 |Body structure (body structure)|   )]], [[~0..1]]   246454002 |Occurrence (attribute)|   = [[ +id(<   282032007 |Periods of life (qualifier value)|   ) @occur]], [[~0..1]]   370135005 |Pathological process (attribute)|   = [[ +id ( <   308489006 |Pathological process (qualifier value)|   ) @proc]], [[~0..1]]   246075003 |Causative agent (attribute)|   = [[ +id (<   410607006 |Organism (organism)|   OR <   105590001 |Substance (substance)|   OR <   78621006 |Physical force (physical force)|  OR <   260787004 |Physical object (physical object)|   ) @causative]] }, [[~1..1]]{ [[~1..1]  363713009 |Has interpretation (attribute)|  =  281300000 |Below reference range (qualifier value)| , [[~1..1]]  363714003 |Interprets (attribute)|  =  441689006 |Measurement of total hemoglobin concentration (procedure)| }

GCI axiom 2:

Definition status:  

900000000000074008 |Primitive|

Template Language

116223007 |Complication (disorder)|  : [[~0..1]]{[[~0..1]]  263502005 |Clinical course (attribute)|   = [[ +id (<   288524001 |Courses (qualifier value)| ) @course]]}, [[~0..1]]{[[~1..1]]  42752001 |Due to (attribute)|  = [[+id (<<  64572001 |Disease (disorder)|  OR <<  71388002 |Procedure (procedure)| ) @dueto ]]}, [[~1..1 @rolegroup]]{ [[~0..1]]   116676008 |Associated morphology (attribute)|   = [[+id(<<  49755003 |Morphologically abnormal structure (morphologic abnormality)|   ) @morphology ]], [[~0..1]]   363698007 |Finding site (attribute)|   = [[+id(<   123037004 |Body structure (body structure)|   )]], [[~0..1]]   246454002 |Occurrence (attribute)|   = [[ +id(<   282032007 |Periods of life (qualifier value)|   ) @occur]], [[~0..1]]   370135005 |Pathological process (attribute)|   = [[ +id ( <   308489006 |Pathological process (qualifier value)|   ) @proc]], [[~0..1]]   246075003 |Causative agent (attribute)|   = [[ +id (<   410607006 |Organism (organism)|   OR <   105590001 |Substance (substance)|   OR <   78621006 |Physical force (physical force)|  OR <   260787004 |Physical object (physical object)|   ) @causative]] }, [[~1..1]]{ [[~1..1]]  363713009 |Has interpretation (attribute)|  =  281300000 |Below reference range (qualifier value)| , [[~1..1]]  363714003 |Interprets (attribute)|  =  14089001 |Red blood cell count (procedure)|  }

Link to the misaligned concept report

Due to Disease -

Due to Disease or Procedure -

Rules for generating descriptions:

  1. Apply General rules for generating descriptions for templates


  1. Bruce Goldberg With the supertype here, shouldn't it be 414022008 |Disorder of cellular component of blood (disorder)|?

  2. Cathy Richardson . Actually, I had meant to delete this template. At the very least this needs more discussion. I based this on the model of anemia that is currently in SNOMED which defines anemia as both a low hemoglobin and an erythropenia. This is in contrast to all of the definitions I can find online which indicate anemia is a low hemoglobin or an erythropenia

    e.g WHO definition: 

    Anaemia is a condition in which the number of red blood cells or the haemoglobin concentration within them is lower than normal.

    I will leave this template up but changed the description to on hold - pending further discussion during an author's call. I believe this topic has been discussed in the past.

  3. Jim Case Bruce Goldberg - I would like to pick this up again as it has appeared as an issue driven by ICD-11 content. I have discussed with Dave Robinson and he is interested in assisting me with this revision. Any objections? thanks, Monica

  4. Monica Harry ,

    No objection from me.  Do we have a way of representing a GCI in a template?  I assume that is what is required here.

    1. That is a good Q... since it is either/or then yes, GCI is what I was thinking...I will check with Yong and Peter. 

  5. I have no objections either if GCIs are indeed allowed in templates. If so, that would help in creating more templates representing disjunctive states. I eagerly await feedback from Peter and Yong.

  6. Yongsheng Gao Peter G. Williams - See previous discussions. Is there a way to represent GCIs in a template?

    1. Hi Monica Harry, yes, we can represent a template for GCI axiom. It is similar to normal axiom but with the heading of GCI axiom. Please see the section GCI model for Structure concept in the template for skin structure of a region at  [Skin] of [body region] (body structure)

      I have added the two axioms in the template and updated the existing axiom, template language and format.

      1. I'm not clear about what a GCI on a template would mean, or how this is different from an additional axiom in which case we could use a single template with 2 x optional interprets/has interpretation groups.   Could we discuss on our Wednesday call?

  7. Thanks both and yes Peter, I was thinking along the same lines of the optional RG. Will review as Yong suggests and def. discuss on Wed. 

  8. The problem with either using GCIs or optional RGs for interpretations of low hemoglobin or erythropenia is that subtypes of anemia would need to be defined as either having either low hemoglobin, low RBCs or both. In most cases of anemia both indices are affected but sometimes only one measurement may be reduced e.g. mild hypochromic anemias not have a low hematocrit and in thalassemia with extreme microcytosis in which the RBC count may be increased. It would be very difficult imo to assign one or more appropriate RGs to the 514 descendants of anemia. I had brought this up during a Tuesday authors call and we agreed at that time that anemia could not be sufficiently defined.

    A possible solution however is based on the “rigorous” definition of anemia from UpToDate below. If we create an observable entity of red cell mass, we could define all anemias with a single role group of interprets red cell mass +has interpretation below reference range.


    Anemia — Anemia can be rigorously defined as a reduced absolute number of circulating red blood cells (ie, a reduced red blood cell mass as determined via blood volume studies). However, blood volume studies are not practical for this purpose, cost-effective, or generally available. As a result, anemia has been defined as a reduction in one or more of the major red blood cell (RBC) measurements obtained as a part of the complete blood count (CBC): hemoglobin concentration, hematocrit (HCT), or RBC count. In practice, however, a low hemoglobin concentration or a low hematocrit is most widely employed for this purpose.

  9. I have added a few comments as this work overlaps with the work I am currently doing on 118245000 |Measurement finding (finding)| in which all of the anaemias currently reside as a consequence of using the INTERPRETS/HAS INTERPRETATION role group.

    I have a number of questions, of which the answers would help me model more consistently the content of 118245000 |Measurement finding (finding)| concepts and may have a bearing on the work planned for the anaemias.

    There are a number of potential duplicates as determined by their modelling in this hierarchy e.g.

    62574001 |Erythropenia (finding)|

    165423001 |Red blood cell count low (finding)|

    Erythropenia is defined as:

    "the presence of decreased numbers of erythrocytes in the blood, as occurs in some forms of anaemia. Also called: erythrocytopenia" -  Steadman's medical dictionary

    Interestingly UpToDate returns no matches on a search for 'erythropenia' and PubMed only 67 many of which relate to animal studies.

    We have said that findings represent a 'point in time' state and so the interpretation of measurement findings should probably be restricted to an interpretation of the single observation and not an ongoing state.

    So my questions are:

    1. Is 'erythropenia' a true synonym of 'low red blood cell count'? - I think this may be unclear, but if it were agreed to be a duplicate it would resolve all of the remaining queries.
    2. Should we have a means of representing the ongoing state, as represented by a number of repeated measurements of low red blood cell counts in the same way as we have 45007003 |Low blood pressure (disorder)| and 12763006 |Finding of decreased blood pressure (finding)| and would 'erythropenia' modelled as a disorder be a way of achieving that? - recognising the many discussions we have had regarding findings v disorders
    3. Should we be using a role group that specifies a point in time measurement to represent an ongoing state, given that it is the ongoing state which often/usually determines the presence of a disorder?
    4. If we wish to retain 62574001 |Erythropenia (finding)| as representing the ongoing state, retained as a finding how would we model the concept to show that it was an ongoing state rather than a point in time interpretation of an observable?

    Opting for the simple solution would be my preference

    1. Paul Amos ,

      62574001 |Erythropenia (finding)| is an interesting concept in that it is modeled as a decreased in red blood cells but has synonyms that refer to both a decrease in haemoglobin and a decrease int eh number of RBCs.  Since "erythro-" simply means "red", the term literally means "red deficiency". I would say it is ambiguous at best and we should be explicit as to whether we are talking about a decreased in the absolute number of cells (i.e. erythrocytopenia) or a decreased in the amount of hemoglobin per cell (hemoglobinemia).

  10. Discussed with Bruce Goldbergand he will raise this at the next QI call for resolution.

    1. Thank you. I intended to raise it myself but had a conflict with last week's call and again now with tomorrow. If we can reach agreement on definition, modeling etc. I can do some of the rework. many thanks, Monica

  11. The GCI axioms represent the disjunctive meaning for either lower haemoglobin, low RBC, or both. It is different from multiple role groups in an axiom. This is also different from multiple axioms. I created a simple Apple ontology to demonstrate the differences in model and classification results. 'Sweet or red apple', as a disjunctive concept, is modeled by two GCIs. You might also notice some interesting side effects of using multiple axioms.

    After looking into a bit more detail of subconcepts of anemia, the GCIs are not necessary for classifying these subconcepts. Most of these subconcepts are belong to association patterns. Please note these GCI axioms are not included in the inferred relationship file. So, they won't facilitate queries based on attributes/values. Given the above the discussion, I think treating anemia as a primitive concept, similar to diabetes, would be okay to model those subconcepts. Let's discuss at the QI call.

     GCI and multiple axioms.owl

    1. I imagine the side effect of using multiple axioms is that all descendants of sweet apple would also inherit the colour red which is not necessarily true?

      1. The side effects are:

        'Sweet red apple' (modeled by multiple equivalent axioms) is a parent concept of 'Sweet green apple'

        and it is also equivalent to a few existing apple concepts.

    2. Yongsheng Gao ,

      I took a look at the subtypes of anemia and found that there are cases of both decreased in cell numbers, e.g. 46737006 |Normocytic normochromic anemia (disorder)| and decrease in hemoglobin, e.g. 44452003 |Normocytic hypochromic anemia (disorder)|, so this seems like a compelling reason to further investigate GCIs for this.  Almost all of the primitive subtypes have a stated parent of Anemia, which is currently SD.  I think that because we have a pretty solid definition of anemia that we can implement as GCI, we should try and take advantage of the benefits.  

      1. Jim Case I find 2 normochromic anemias in SNOMED. As these are anemias with normal  hemoglobin , they must have low RBC numbers. There are 4 hypochromic anemias. RBC numbers in these may or may not be decreased depending on the stage of the anemia. There are likely other specific anemias that would classify as being hypochromic and/or having decresed RBCs but this would take  a lot of work to identiify. Are you sure it is worth the effort to create a GCI model for anemia?

        1. Bruce Goldberg ,

          There are over 500 subtypes of 271737000 |Anemia (disorder)| currently.  Many of these have 271737000 |Anemia (disorder)| as a stated parent, which may indicate that it is the presence of the word "anemia" that determines whether a concept becomes a child.  My impression in creating a GCI modeled Anemia concept is that there are concepts in the terminology that are not properly classified as anemia because they do not "have the word" in them.  This is all supposition, but given the tools we have, I think we should investigate leveraging them to improve the inferences.  

          If we do not use GCIs, then we must make the current Anemia term primitive as it currently requires that both decreased numbers AND decreased hemoglobin be present. The disadvantage of the primitive concept is that it must be manually curated.  The disadvantage of the GCI based term is the amount of research that needs to be done to ensure appropriate classification.  

          Open to other opinions...

          1. Hi Jim Case Bruce Goldberg the model of two role groups is incorrect in the current release because it means that both conditions (low haemoglobin, low RBC) are essentially required for defining anemia. As discussed, this is an incorrect representation for disjunction. Therefore, anemia can only be a primitive concept. From the authoring perspective, the concept 'anemia' is the obvious primitive parent for concepts have the word 'anemia' in their descriptions. One option could be that we only add the GCI axioms for 271737000 |Anemia (disorder)|. This will help capture the subtypes that do not have the keyword 'anemia' by concept model. However, it is not necessary to add GCIs to subconcepts of 271737000 |Anemia (disorder)| because it does not add any obvious additional benefits for classification and queries. This could be a balanced approach. Thoughts?

            1. Yongsheng Gao ,

              I agree that we should test the GCI approach.  My only caveat would be that we would need to add GCIs for the concepts that Bruce pointed out where there are subtypes that also represent and/or decreased numbers or hemoglobin.  Agree that GCIs would only be needed for a very few concepts such as the parent or other potentially disjoint concepts.

              So, what we need to decide is which of the two options; Anemia as a primitive or Anemia as a GCI modeled concept provides us with the best long-term solution.

  12. Jim Case , Yongsheng Gao  As an example of a concept which is currently classified under anemia but does not contain "anemia" in the FSN there is 19442009 |Heterozygous thalassemia (disorder)|. It is modeled as shown with:

    363713009 |Has interpretation (attribute)| = 281300000 |Below reference range (qualifier value)|,
                  363714003 |Interprets (attribute)| = 14089001 |Red blood cell count (procedure)| }

    which would classify this concept under anemia. In truth however the RBC count may be increased in Heterozygous thalassemia (Thalassemia minor). A GCI would still classify this under anemia if the the total hemoglobin is low. The MCH is low in the Thalassemias but I need to ascertain whether total hemoglobin is also low.

  13. Thalassemia minor is characterized by reduced MCV and MCH, with increased Hb A2 level [29].

    Galanello R, Origa R. Beta-thalassemia. Orphanet J Rare Dis. 2010;5:11. Published 2010 May 21. doi:10.1186/1750-1172-5-11

  14. Hi Bruce Goldberg, because anemia has been incorrectly defined by two role groups, all subconcepts of anemia have inherited lower RBC and lower hemoglobin. We cannot rely on the inferred model of subconcepts of anemia. Therefore, each model of its subconcepts has to be reviewed and remodeled. 

    "Two separate and partially overlapping types of anemia: hypochromic anemias, resulting from a decreased amount of hemoglobin in red blood cells, and hemolytic anemias, resulting from increased red blood cell destruction". "The combination of hypochromia and increased hemolysis defines a distinct subset of entities, of which thalassemias are the most common". Atlas of Diagnostic Hematology Copyright © 2021 by Elsevier, Inc. All rights reserved.!/content/book/3-s2.0-B9780323567381000038?scrollTo=%23hl0001043

    The concept  40108008 |Thalassemia (disorder)| can be model by low hemoglobin as a necessary condition. Then, it will be classified under the GCI anemia.  The due to relationship also needs to be modelled in order to be classified under 14514008 |Anemia due to disturbance of hemoglobin synthesis (disorder)|

    Thalassemia is anemia of varying severity due to inadequate production of globin chains
    Thalassemia is a quantitative hemoglobin disease; hemoglobin is normal but an insufficient amount is produced

    Elsevier Point of Care (see details)
    Updated May 7, 2019. Copyright Elsevier BV. All rights reserved.!/content/clinical_overview/67-s2.0-bed2df48-2c67-43b5-b2b6-40e4e2a22571

    More detail about the classification of thalassemia can be found in the above two references as well. 

    1. Hi Yongsheng Gao. I understand that any disorder modeled with low hemoglobin, low RBC count or both would classify under Anemia if Anemia is modeled using 2 GCI axioms.  What is the due to relationship that would need to be modeled for Thalassemia to be classified under 4514008 |Anemia due to disturbance of hemoglobin synthesis (disorder)|? This is currently a primitive concept with no defining relationships. Are you saying we would need to create and define something like disturbance of hemoglobin synthesis (disorder)|? 

      Some additional issues with Thalassemia: 

      1. Thalassemia can also be considered to be a hemolytic anemia and an abnormal erythropoiesis.
      2. The RBC count is normal or increased in Thalassemia. If we could represent "normal or increased" then I think Thalassemia would no longer classify under anemia.

      So when you say "Therefore, each model of its subconcepts has to be reviewed and remodeled.", this will be a huge effort. There are 71 descendants of anemia without "anemia" in the FSN

  15. Bruce Goldberg,

    At this point, since 14514008 |Anemia due to disturbance of hemoglobin synthesis (disorder)| is primitive (as are all of its subtypes), could we not just use that as the parent for Thalassemia until such  time as we can decide whether this term is needed or can be defined? It would still classify under Anemia.  It would not conform to this template, and we could create a new one to support these concepts.

    1. Jim Case,  I will create a template for Thalassemia which I think will be complex due to issues 1 and 2 I mention in my Jan. 16 post. 

      Likely we will have many concepts that won't conform to the GCI model for anemia. 

      1. Hi, Bruce Goldberg, there are 66 concepts under 40108008|Thalassemia (disorder)| among those 71 descendants of anemia without 'anemia' in the FSN. After Thalassemia hierarchy is remodelled, there are only 5 concepts that need to be reviewed and remodelled. The most effort would be the remodel of Thalassemia as you are proposed here. 

        The RBC count normal or increased in Thalassemia won't stop Thalassema be classified under anemia. The low RBC count is not a necessary condition for anemia but one of the sufficient conditions. So, Thalassemia would still be classified under anemia as long as it has a necessary condition of low hemoglobin. The classification has been demonstrated in the Apple ontology. 'Red or sweet apple' modelled by GCIs would still subsume 'Green sweet apple' because it is 'sweet' (smile)  Normal or high RBC count is similar to the 'Green' color is not a subconcept of 'Red'. That won't stop the subsumption by sweetness in GCI.

        I agree with Jim Case, the intermedia parent with due to could be addressed later. 

  16. Yongsheng Gao,

    Thanks for updating this for the GCI, but I am wondering what the need for all of the attributes are in the top axiom as all of the relationships are optional. Could the top level axiom just be a subtype axiom for Disease? Also, you have the subtype axiom as 0..1 as well.  Is that intentional?

    1. Hi Jim Case, good point. I think we can archive this template after we agreed on the approach. There won't be attribute/values for the normal subclass axiom for this concept.  The new template should focus on the model of its subconcepts such as 'Anemia due to X. 

  17. I agree and will create a new template at the end of this week.

  18. See Anemia due to [Clinical finding/Disease] - revised template - for review and Thalassemia - new template - for review under Association templates

  19. Remodelled Anemia as per template. 

    << 414022008 |Disorder of cellular component of blood (disorder)|: {363713009|Has interpretation (attribute)| = 281300000|Below reference range (qualifier value)| and 363714003 |Interprets (attribute)| = 441689006 |Measurement of total hemoglobin concentration (procedure)|} {363713009|Has interpretation (attribute)| = 281300000|Below reference range (qualifier value)| and 14089001 |Red blood cell count (procedure)|}

    Filter by term = Anemia 
    and "State" with definition type = All
    The 'State' will exclude those concepts inferred two RGs from |Anemia|. Those 'Inferred' should have already lost 2 RGs after the model change to |Anemia|.

    A simple example, 413532003 |Anemia due to blood loss (disorder)| can be remodelled as fully defined by:  |Anemia| is the PPP parent.  
    IS A = |Anemia| and |Due to| = |Bleeding|
    This type of concepts does not need to have two GCI axioms.