OHDSI Home | Forums | Wiki | Github

AMT- RxNorm mapping source?

Hi Everyone,

It’s just a quick question on where is the AMT - > RxNorm mapping sourced from, or is the mapping done by OHDSI.

Cheers,
Guan

Hi Guan,
yes, it’s done by OHDSI Vocabulary Team.
Do you have any additional questions? Feel free to ask!

@aostropolets

Thanks Anna.

I also want to know whether there are any existing AMT -> Rxnorm or AMT -> ATC mappings other than these done by OHDSI.

Existing - meaning done by other people? Do not think they exist.
AMT have relationships to SNOMED substances, some of which have mappings to RxNorm ingredients.
Why? Do you have a specific use case?

Thanks Anna.

To put it in context, we have found another AMT to ATC mapping done by PBS in Australia. However, these we don’t exactly know how they mapped it and we would assume that being a medication reimbursement scheme, the mappings are done for their purpose, we are interested in how accurate and reliable these mappings are in comparison to whats done in OMOP for our purpose (drug utilisation/cohort identification).

I have a few questions and follow ups from previous comments

some SNOMED substances have existing mappings to RxNorm ingredients.

From where? UMLS?

So mappings only exsist at the substance level.

What about other levels? trade product, trade product unit etc. Were these all matched one by one by the volcabulary mapping team manually?

RxNorm is partially sourced from ATC, so the mapping is already existing between RxNorm and ATC. and this is only at the ingredients level I’m guessing? And and ATC classification is linked to individual drugs based on RxNorm internal links between ingredients and drugs. is this correct?

And does anyone have the numbers of how many RxNorm were mapped to ATC and how many of AMT was mapped to ATC through RxNorm in OMOP.

correct

yes, but not manually, but using scripts for a new vocabulary creation.

yes, all correct.

Please send your file with AMT-ATC mapping. We will greatly appreciate it! :smile: Then we can test them and say whether they are usable.

Thanks Anna, these are very insightful information.
May I ask how are you going to test the usability and accuracy of the mappings? and what measures you plan on using?

Great question! The first thing that comes to my mind is to compare the mappings to the existing ones. Then check if ATC5th that have underlying dose forms are mapped from the drugs that have the same forms.
Maybe you have other use cases to check?

Can I ask for your email so I can send through the mappings from PBS?

With the SNOMED substance -> RxNorm mapping mentioned earlier. At the moment I believe AMT and SNOMED substance are still separate entities, even though they are equivalents. how are they joined up?

Also with the mappings of AMT (trade products packs, etc) to RxNorm mappings. You mentioned they were done with a script. Would we be able to take a look at the script to understand how it is matching the terms?

Thanks so much.

Hi there Anna,

can i possibly ask please whether there is any further work or checks on the PBS medication linkage …
we have a large data set comprised of medications data with australian PBS item codes … and we would love to be able to get the data into the ohdsi CDM form … any help would be appreciated …

maybe can we correspond by email (andrew.milburn@unisa.edu.au)
kind regards,
andrew

Hi Andrew,

With a medication database with PBS item codes, you can use the PBS to AMT mappings that PBS has provided on their website. You then can load the AMT codes into OMOP CDM and have it converted to the standard concept via OMOP AMT-> RxNorm (standard concept) mappings.

We have just been working on converting a database of EHR data into OMOP CDM. I’m happy to answer further questions.

You can get in touch with me (guan.guo@unsw.edu.au) or Prof.Teng Liaw (siaw@unsw.edu.au)

Cheers,
Guan (Jason)

@aostropolets

Hi Anna,

In regarding the mapping by OHDSI between AMT (substance) -> SNOMED (substance) there are still a few areas I’m not 100% sure about. AMT and SNOMED - CT have large overlaps at the substance level. However, they are not systematically connected (linked or mapped). There is still the question on how were SNOMED-CT substances linked with AMT substances (MP). You mentioned that SNOMED-CT UK has mapping to AMT at the substance level through dm+d, how do we retrieve this or look at this mapping? I’m not able to find anything here (https://apps.nhsbsa.nhs.uk/DMDBrowser/DMDBrowser.do#product)

Cheers,
Guan.

Guan,

AMT and SNOMED UK are the subsets of SNOMED. That’s why they have an overlap in codes. And sorces provide us with the links (like https://www.digitalhealth.gov.au in this case).

The other problem is this: SNOMED does not do an adequate job of incorporating drugs. In fact, they want to drop all the products and just leave the active compounds in the future. In addition to that, they have the two categories “Substance” and “Biological or Pharmaceutical Product”. Both are overlapping, and that only partially. Substance seems to mean any substance, whether it came into the body through drug administration or another route. The whole thing is not clean.

Why do you want to create this mapping anyway? What’s the use case?

Thanks Christian and Anna

My purpose for asking this isn’t to create such mapping, but understanding how AMT was mapped to RxNorm (existing concepts) and then to ATC codes in OMOP CDM.
Really my big question is to know the steps on how AMT was mapped to RxNorm and what existing mapping sources were used for the incorporation of AMT into the CDM.

t