Terminology: Does the abdomen include the pelvis?

Good afternoon everyone,

SNOMED International has responded to the feedback we sent regarding the abdomen and pelvis discussion as part of this thread. Here is their response:

Thank you for your email with the subject "Canadian DI implementers and concerns about recent changes: Abdomen and Pelvis".

These changes are part of the anatomy revision. The intention was to make a clear distinction between "Abdominopelvic structure" and "Abdomen proper (Abdomen excluding pelvis) structure". We also received comments from the DICOM. Following the feedback, we are reviewing the changes in the release and will produce a document for concepts related to the abdomen and their definitions. We would appreciate if you can help to review the document and get the input from the clinical experts.

I am happy to arrange a call to update the current progress if you think it is useful.

Kind regards,

Yong
Yongsheng Gao

Please do let me know if there is interest from the community in having a call with SNOMED International (SI) to discuss as suggested above.
If not, we will definitely be looking for members of this community and the Health Terminologies community to review SI document and provide feedback when it is ready.

Thank you.

Regards,

Andrea

Thank you David and Cezary for reaching out and raising awareness of this change and concern. I have shared this message thread, along with the feedback and examples collected from the following individuals/organizations with SNOMED International:

• eHO
• Dr. David Koff, who surveyed all of the academic radiology departments in Canada
• BC MoH
• David Kwan, now of CCO who helped develop the original DI content with eHO

Thank you all. It is appreciated.

If there is other feedback or uses cases that you would like to share pertaining to this thread, please feel free to do so. We are happy to pass it to SNOMED International.

We have requested a follow up discussion to learn more, so stay tuned.

Thank you.

Regards,

Andrea

Hi David,

I think your concern is justified: what SNOMED has just done is a rather frivolous semantic adjustment for the meaning of 113345001. It has been a common understanding, also in Canada, that abdomen is separate from pelvis and they are both children of "abdomen and pelvis". This indeed seems like a tough one for the DICOM standard, with no obvious path out of this, unless you resort to updating all prior instances (not sensible/possible). I am puzzled by the SNOMED decision, to say the least. This results in a BREAKING CHANGE for those in the DICOM world who naively though that "abdomen" meant "abdomen" ?!. I hope SNOMED-CT gives that breaking change some consideration as this is certainly not the greatest marketing campaign they've embarked on.

In Canada (or to be more specific in BC), we have always assumed that procedure codes would change at some point (including breaking changes), or the semantics of certain codes would change. Consequently, in BC, we decided to take an approach where we map imaging procedures codes coming from local enterprises to the provincial set (and back) and we do not rely on what has been embedded in the datasets. We have realized that imaging terminologies cannot be considered on their own merit but rather in the context of the environment where they are being used.

As a precaution and insurance against "floating" imaging terminologies, we have resorted to explicit terminology mappings between LOCAL and PROVINCIAL sets. For example, if a local enterprise were to follow the lead of SNOMED-CT and start treating the code "CT of abdomen" to mean "CT of abdomen and pelvis" we would expect the chief radiologist there (or whoever is in charge of mappings) to adjust the mappings between this enterprise's terminology and the provincial terminology to "CT of abdomen" [ENTERPRISE] --> "CT of abdomen and pelvis" [PROVINCE]. The reverse mappings would also need to be adjusted accordingly so when we brought in a foreign prior coded "CT of abdomen and pelvis" [PROVINCE] it would end up as "CT of abdomen" [ENTERPRISE]. Some other institution could be different. Similarly, if the meaning of a procedure code were to change at the provincial level, e.g., "CT of abdomen" [PROVINCE] were to start meaning "CT of abdomen and pelvis", we would need to adjust the affected translation tables to/from local enterprises.

The above decoupling of local and global/provincial terminologies, combined with the management of codes (procedures, body parts, coarse body parts, modalities, actionable findings, clinical findings, etc) inside a metadata registry allows us to:
1) maintain a consistent meaning of all codes at the provincial level even when a contributing enterprise that happens to use the same terminology has a different meaning for that code
2) perform the search for relevant priors against the metadata registry using the authoritative/PROVINCIAL set of codes (with authoritative semantics) rather than against the codes and semantics that are enterprise-specific and time-specific (this year's or last year's meaning?) which end up embedded in the DICOM datasets.

Cezary

P.S. In BC we are not using SNOMED-CT for identifying imaging procedures at the provincial level, but the problem of "floating semantics of codes" is occurring in other terminology universes as well, including those home-grown. The mapping approach insulates us from this problem. The province decided against forcing a single imaging terminology onto all healthcare enterprises there due to viability concerns.

SNOMED CT has recently been doing major work on cleaning up their anatomy.

One of the changes is to redefine the current code for "abdomen" (113345001) to mean "abdominopelvic", i.e., to include the pelvis.

This has procedure code consequences, since:

CT of abdomen
CT of abdomen and pelvis

will now mean the same thing, semantically

SNOMED has argued that clinically "abdomen" has a different meaning than cross-sectional radiologists might use it (to not include pelvis), and also points to the plain X-Ray of the "abdomen" as including the pelvis. and it is hard to argue with that, but they are reluctant to "inactivate" 113345001 and have retired the current "abdomen and pelvis" code as being a duplicate There is an "abdomen proper" code added.

This also invalidates all DICOM images that have previously used the SNOMED code for abdomen to distinguish something from pelvis, or abdomen and pelvis. (and I am supposed to update the DICOM standard to follow the new SNOMED model).

There is no obvious pathway out of this dilemma, but I wanted to get the Canadian radiology take on how you think this impacts the procedure codes and hanging protocols and prefetching rules, since Canada is one of the few countries using SNOMED procedure codes. I will ask the UK folks the same question.

I also want to confirm that this is a genuine concern and that I am not overreacting, as well as to ascertain when (for which modalities and procedures) radiologists mean "abdomen and pelvis" when they say "abdomen" and when they mean "abdomen without pelvis".

The editorial guide for the procedures in Ontario seems to be pretty specific about using Abdomen separate from pelvis.

David

PS. I argue that the class hierarchy in SNOMED has always separated the codes for abdomen from pelvis, being siblings, and both children of (to be inactivated) "abdomen and pelvis", but SNOMED argues that their editorial guide describes abdomen from the frontal clinical exam perspective, and that they have used the latter to resolve the conflict and justified renaming the existing code.

PPS. Sorry I won't be able to make your FEM call this morning, though this topic would seem to be relevant.