OHDSI Home | Forums | Wiki | Github

ICD-O-3 Histology Concept Class

(JD Liddil) #1

So I downloaded the ICD-O-3 and find the the Histology Concept Class is not present specifically:
B lymphoblastic leukemia/lymphoma with t(v;11q23); MLL rearranged

  • Domain IDObservation
  • Concept Class IDICDO Histology
  • Vocabulary ID ICDO3
  • Concept ID44498906
  • Concept code9813/3

Why not?

(Christian Reich) #2

Not sure I understand. What is it you don’t find? The concept is there, and the concept class is ICDO Histology. What’s wrong?

(JD Liddil) #3

I find this information in Athena but not in the vocabulary using Usagi. I should have been mpre specific.

(Dmytry Dymshyts) #4

Have you checked the ICDO3?

image
image.png949x95 4.05 KB

(JD Liddil) #5

Yes. I guess they don’t show in Usagi since they are non-standard?

(Christian Reich) #6

ICDO3 is implemented with three concept_class_id:

  • ICDO Histology
  • ICDO Topography
  • ICDO Condition (which is a Cartesian product between Histology and Topography, but only those combos that exist in real data are instantiated in the vocabulary).

Only Conditions are Standard Concepts. So, for histology and topography you need to pick SNOMED instead.

@Dymshyts: Why did we make Histology and Topology not standard, if we don’t have a SNOMED equivalent? Do you remember?

(JD Liddil) #7

Well in our case (cancer research drug development) we confirm both Topography and Histology and check that they agree. Either way we gather both data points. Topography is generally confirmed by a clinician and the Histology is a pathological confirmation. SNOMED is often inadequate for this vs the ICD coding.

(Christian Reich) #8

Agree? Like the histology can happen in the topology? We did that with the ICDO Condition. So, a melanoma of the skin cannot have adenomcarcinoma as the histology.

Both data points are present in the Condition. Pick the ICDO3 histology code, the ICDO3 topology code, combine them with a dash and look up the chimera. For example, you got ICDO3 9700/3 “Mucosis fungoides” and C42.0 “Blood”. Look up 9700/3-C42.0 and you will find Condition Concept 36402813 “Mucosis fungoides of blood”.

(JD Liddil) #9

So back to my specific example. I was obviously not looking hard enough
https://phinvads.cdc.gov/vads/ViewCodeSystemConcept.action?oid=2.16.840.1.113883.6.96&code=450951003

(Christian Reich) #10

Tons of them. Check out Athena. But you need to combine it with the topology. Where does the lymphoma sit? Without it, it’s just a histological finding, not a Condition.

(JD Liddil) #11

I use Athena but it does not always seem to find things. Again I don’t write the questions. So “we” don’t ask anything more specific about the site in this case. Part of the issue faced is some of this is legacy data we are trying to analyze/mine. I’m faced with trying to doing this exercise with folks that don’t have a clear idea about the use cases. :slight_smile:

(Christian Reich) #12

Well, in this case you can probably default to “blood”, if you cannot find affected lymph node information in the data.

What data is it?

(Dmytry Dymshyts) #13

this particular concept has SNOMED equivalent

image
image.png1032x46 2.7 KB

But there are some other non-standard and without mapping to SNOMED.
So, the answer why - just algorithm flaw I think:)
I fix it.

Here are the other fixes planned:

  1. add “Is a” relationships between ICDO and SNOMED attributes.
    currently they are not added,
    ‘Is a’ between ICDO and SNOMED attributes.
  2. change ‘ICDO - SNOMED’ relationship to ‘Maps to’. It’s a basic ‘Maps to’ relationship - non-standard concept is mapped to standard
  3. take these changes into account - rename concepts
    image
    image.png1056x261 21.4 KB
  4. fix valid_end_date for concept_relationships, use current date instead of valid_start_date used now
  5. 8502/3-C50.9 “Secretory carcinoma of breast of breast”-like cases
    And I really like this discussion. At last we have ICDO tested by the community.
(JD Liddil) #14

Cool. Thanks. Yes I’ll be using ICD codes quite a bit and giving feedback.

(Christian Reich) #15

Why is this a flaw? If we have a SNOMED equivalent we use that. If not, we use ICDO3, correct? What are the rules for standard concepts?

Are there new in ICDO3?

We will have a lot more of that in the Oncology WG

(Dmytry Dymshyts) #16

It’s not an algorithm flaw, right. We decided that ICDO3 attributes (Topography and histology) are always non-standard regerdless of presense of mapping to SNOMED.
While ICDO3 condition concepts are not standard if only the have SNOMED equivalent.

Now I think that the rule for ICDO3 conditions should be applied to Topography and histology as well.

9086/3-C73.9 is new, 8343/3-C73.9 already exists.

Then we need to use new names, because “bronchioloalveolar carcinoma” is no longer used (the codes contained this name).
Note, the clinical meaning stays the same. Don’t worry, they don’t make “concept changing their meaning over time” problem.

(Christian Reich) #17

Me too

Yes, if they change the name we have to change it as well.

t