OHDSI Home | Forums | Wiki | Github

Difference between vocabulary ICD10 and ICD10CM

Vocabulary Users
#13

In the vocabulary files I downloaded on 2016-07-13, the concept_relationship table contains hierarchical relationships for ICD10, but not ICD10CM. Are the hierarchical relationships for ICD10CM available? Or were they just not built / added to concept_relationship?

EG: ICD10
“45571628 Iron deficiency anemia D50” Subsumes “45605373 Iron deficiency anemia, unspecified D50.9”

In ICD10CM, the concepts for concept_codes D50 and D50.9 are “1567838 Iron deficiency anemia”
and “35206695 - Iron deficiency anemia, unspecified.”

But the only relationships present in concept_relationship for these ICD10CM concepts is the mapping to/from SNOMED.

#14

@donohara:

You are right. We will add these in the next release.

But why do you need them? The ICD10CM hierarchy leaves much to be desired: It is realized through “microcoding” of the identifiers (bad because if an identifier is changed for some reason the hierarchy has to change as well), it allows only one parent, and it only has a few levels in depth. I wouldn’t use it. Use SNOMED.

#15

Thanks @Christian_Reich. Do you know when the next release will be? Our clients still use the “hierarchy”, warts and all. We have talked with them about its deficiencies, and the long term goal is to move to using SNOMED, but in the short term the hierarchy is required.

#16

This week.

#17

Hi @Christian_Reich: An update on the ETA for the ICD10CM and ICD10PCS hierarchical relationships? The latest vocab download does not have them.

1 Like
#18

+1 for the ICD10PCS hierarchy.

#19

Hi @Christian - Just curious if you have an update on when the hierarchical relationships will be available for ICD10CM and ICD10PCS ?

#20

@Dymshyts:

Did you guys put them into the last release?

#21

Yes, we did

#22

@Christian_Reich, if you mean relationships to SNOMED, so we didn’t.
We just have hierarchical relationships between ICD10PCS themselves, for example
DT1297 Subsumes DT1297Z
BB0 Subsumes BB08

#23

Hello,

Can anyone let me know that what is “Concept Type Vocabularies”.

where can i find theses vocab as theses are not available at Athena .

I would also to know that how to use and it would be really appreciable if someone can quote some examples.

Is there any vocabulary for location as well??

Thank you.

#24
#25

Is there a plan, work in progress to come up with a single vocabulary hierarchy? Or is one forever cursed with dealing with multiple vocabularies and having to reconcile them?

#26

@optimizer:

What do you mean by “vocabulary hierarchy” or “reconciling” vocabularies? ICD-10 is a product of the WHO, and ICD-10-CM a derivative maintained by the CMS, which is part of the US government. Other countries also have their own national versions of ICD-10. These guys just do these things, there is not a lot we can do about that.

#27

An oversimplified example would be character sets. We started with ASCII and now have character sets that can represent any language.

The actual physical conditions, drugs, procedures are the same regardless which vocabulary is used to reference them, so there must be a way to homogenize the codes and therefore the mappings from terms to codes.

Yes, I get your point that different agencies/countries just do their own stuff without regard for consistency.

The motivation for consistency, however, would be to enable building cohorts based on multi-national, multi-agency, multi-institutional data.

#28

@optimizer:

You are totally right. That needs to happen. In fact, we have the explicit goal to have a concept for every medical entity there is, and to have only one such standard concept. For Drugs this is more or less already achieved, for Conditions it’s in a good shape (SNOMED does the de-duping work for us), for Measurements we have LOINC doing the same thing, but Devices, Procedures and Observations need work. Want to lend a hand? :smile:

#29

Although I don’t mind doing some work, I’d prefer to not formally commit to it, as I don’t know up front what projects I will be doing.

That said, I’d propose that this should be a crowdsourcing effort similar to Wikipedia, so that multiple people can adjust, modify, add entries in a spreadsheet (documenting rationales for their edits.) Of course with multiple people being able to edit, one needs some version control mechanism. Github could be used? I am not sure if Google docs allow version control or not.

You mentioned several domains that are in reasonable shape, assuming those data are available in machine consumable format, they can be used to initialize the spreadsheets.

For procedures, devices, observations people in the know could probably identify the most consistent source with most coverage to get the ball rolling.

#30

@optimizer:

But that’s how it works here: You collect a bunch of like-minded people and start working in a shared environment. Apart from Maura, the PM, all of us have day jobs, which sometimes have the tendency to distract us from the fun, here. :slight_smile: So, feel free to chime right in. You seem to have a lot of ideas.

#31

@Christian_Reich, could you explain in more detail what your effort (the one to be finished in January) is?

Are you mapping ICD-10-CM (at least the 50k ones that are only in ICD-10-CM) codes to ICD-10 codes?

Is it text similarity based? Is it being done manually?

#32

All roads are leading to Rome, meaning, everything gets mapped to standard concepts. For conditions (ICDs have mostly conditions, but note the “mostly”) those are derived from SNOMED, and most of them are “Clinical findings”. So, ICD-10 and ICD-10-CM is going to be mapped to SNOMED. But: We need to make sure that the mapping is consistent. Not necessarily identical (the detailed meaning of the same code in ICD-10 and ICD-10-CM might differ slightly), but consistent.

t