How to handle mapping for "non-standard" parent but "standard" child concept_id?

Vocabulary Users
1

Dear All,
I have a concept (behavior code icd-o-3) which maps to this: 914042 (Athena)

This concept has the relationship “Has Answer” for “Benign”, “In situ”, “Malignant primary” and “Uncertain Benign/Malig”.
All the answers are “valid and standard” concepts. But the parent (914042) is non-standard.

How can I represent this in the “Observation” table, since the parent (914042) is non-standard?

Thanks a lot.

2

In OMOP ICDO3 behavior codes are coupled to the morphology codes, and participate in the Standard hierarchy this way.

Assuming you are trying to encode some oncology diagnosis from a question-answer pair, you should instead extract values for the morphology, behavior and topography, and make an entry to the condition_occurence table using ICDO3 concept or it’s mapping target (e.g. Athena)

Note that in lesser amount of cases you may encounter combination for which no concept exist (yet?). Depending on your study intent, you may choose to represent those concepts with partial detail:

https://athena.ohdsi.org/search-terms/terms/36402473
https://athena.ohdsi.org/search-terms/terms/36402365

3

Thank you Eduard for your reply and information. That’s helpful.

In my source data, I actually have the topology, histology and behavior as a separate field. So, what I have done is:

  1. Form the condition from the histology/behavior-topology and put this in the condition_occurrence table.
  2. Capture the histology, morphology and topology from the source data in the OMOP CDM observation table.

This is my first time doing this for oncology data, so I am not sure if this is the way to go and if there is any best practice in this case?

Thanks a lot.

4

Hello @wint,

You should join @agolozar and the Oncology WG. They will guide you on best practices and mapping your data in a standardized way. You can find them on the Oncology MS Teams channel.

5

Hi @wint: You are close. :slight_smile:

With respect to 2: Yes, do that. We do have a little problem that the topography concepts are mostly of the domain “Spec Anatomical Site”, and they would technically belong in the SPECIMEN table. But putting it into OBSERVATION is legal and that’s what we would recommend.

With respect to 1: Yes, we are well aware of this. Here is the problem and the solution:

Cancer conditions are combinations of histologies/behaviors and topographies, for example Squamous cell carcinoma of lung. But SNOMED has so-called “one-legged” condition concepts lacking the topography, such as Squamous cell carcinoma, which you seem to be picking up. Also, similarly, there are conditions containing proper topographies, but the histology is kind of trivial: Malignant neoplasm of lung. We call them “shallow”, and they are a typical product of ICD10 mappings. However, neither one-leggeds nor shallows exist for all histologies or topographies, but more importantly, they don’t help our analytical use cases.

What you need to do is to combine the histology Squamous cell carcinoma or the one-legged condition Squamous cell carcinoma (they have the exact same names, I know) with the topography Lung structure or to the shallow Malignant neoplasm of lung to get to Squamous cell carcinoma of lung.

In ICDO3, the equivalent combination of histologies and topographies is easy - it doesn’t have one-leggeds and shallows and all you do is to concatenate the codes. In SNOMED, it is not. There are also ICDO3-SNOMED chimeras. We can provide you with a combination script that does all that for you (and we are also planning on publishing it, but as usual, we are behind on documentation). So, as Melanie said, come to the Oncology WG and we will help you out.

Regarding the Carcinofibroma of unknown primary site: do you want me to write another page, or want to come to the Oncology WG? :slight_smile:

Regarding the Unknown histology of lung: ditto?

6

Thank you @Christian_Reich and @MPhilofsky. Very much appreciated for the elaborated answer. I will join the Oncology WG for help. That will be my destination. :smiley:. Thanks a lot.

1 Like