OHDSI Home | Forums | Wiki | Github

Ingredient to ATC relationships

Vocabulary Users
#1

I have a terminology relationship question.

It is best demonstrated by an example

Neomycin (athena CID: 21605153) is member of several ATC classes
(such as S02AA07,D06AX04,S01AA03,A07AA01)

This is reflected by relationships with ID ‘ATC - RxNorm’
cid1,cid2,relationship
21603795 1378509 ATC - RxNorm
21603016 1733765 ATC - RxNorm
21602252 1776430 ATC - RxNorm
21605128 19008339 ATC - RxNorm
21601686 1319998 ATC - RxNorm

This is great knowledge in CDM vocabulary. (and great relationship ID)
I am a grateful for such mapping.

Now…
Some ingredient concepts are now in RxNorm Extension.
For example

CONCEPT_ID,CONCEPT_NAME
21014008,Pixantrone
21014017,Agomelatine
21014035,Mifamurtide
21014047,Corifollitropin alfa
21014072,Epoetin theta
21014111,Catumaxomab
21014118,Gimeracil
21014127,Dapoxetine

To map RxNorm Extension ingredients to their ATC class - what relationship can be used?
(for eample for last concept of dapoxetine - the answer should be G04BX14 )

(I suspect the answer will be: no volunteer did that yet. (and that is OK answer. just confirmation of that is perfectly fine answer)

(I prefer to use relationships (and not concept_ancestor table, but solution using that is fine too)

p.s.
current relationships for 21014127,Dapoxetine have no similar relationship in the style of “ATC - RxNorm Extension”.

What I see is

  CONCEPT_ID_1 CONCEPT_ID_2 RELATIONSHIP_ID
1     21014127     21015690  Has brand name
2     21014127     21066484   RxNorm ing of
3     21014127     21105795   RxNorm ing of
4     21014127     21086130   RxNorm ing of
5     21014127     21014127         Maps to
6     21014127     21014127     Mapped from

I am using version 24-APR-17.

(on a related note, our public server seems not to have RxNorm extension concepts at all)
http://www.ohdsi.org/web/atlas/#/concept/21014127)

Starting ATC Mapping Working Group
#2

@Vojtech_Huser:

I am not even going to read what you are saying, because I am sure it makes sense. But: We are doing a big surgery on the ATC right now. Let’s take a look after the next release (this month). It should behave the way you expect it to.

#3

@Christian_Reich Is the new release out? Can we re-open this question.

#4

Coming. ATC is fixed. We are still releasing RxNorm Extension.

#5

@Christian_Reich Is there any “good” documentation on the relationships? There are some that look like redundant (eg, in case of ingredients: Mapped from and RxNorm - ATC), but I believe they don’t. So, having more explanation (clarification) on them is much appreciated!

1 Like
#6

Hi, @abedtash_hamed
it’s a very good point.

Unfortunately the vocabulary team is overbooked now and we’ll make the nice documentation some later time.

#7

@Dymshyts That would be great! For the short term, is there any best practice guideline to use the relationships available somewhere? Just key vocabularies and relationships.

1 Like
#8

@Christian_Reich @Dymshyts I tried another way to map drug concepts to ATC 5th level using ‘Mapped from’ and ‘Drug has drug class’ relationships. I noticed mappings are missing for a good chunk of products, for example insulins. I think we need to fix ATC relationships systematically.

An alternative might be mapping drug concepts to ingredients then ATC 5th level, but this does not give you the correct mappings as one ingredient can be under multiple classes depending on route of administration, dose from, DDD, etc.

3 Likes
#9

Ok. Will start working on this the next week

#10

@Dymshyts Thank you! Let me know if I can be any help.

#11

@Dymshyts Thank you! Let me know if you needed my help.

#12

Thank you too.
I’ll send you the e-mail with my questions

#13

Friends:

Bring in Olivier Bodenreider from NLM (obodenreider@mail.nih.gov). He also has a solution. Let me know if you need an intro.

#14

Yes, I need. Can you please do it

t