OHDSI Home | Forums | Wiki | Github

CONCEPT_ANCESTOR - Ingredients mapping to themselves?

Vocabulary Users
#1

In the process of QCing LAERTES, specifically was looking at drug concepts that did not get mapped to an ingredient.

  • Original LAERTES Vocab Version: OMOP Vocabulary v5.0 17-JUL-15
  • Updated LAERTES Vocab Version : v5.0 11-MAR-16

We use the CONCEPT_ANCESTOR table to get drugs to their INGREDIENT level. There seems to be cases where things used to map to themselves using CONCEPT_ANCESTOR but now do not (see query in APPENDIX A and results below).

Is this a bug or a misunderstanding on my part?

Concepts that No Longer Map To Themselves in CONCEPT_ANCESTOR
19023420 Bile fluid 19080435 FACTOR IX,RECOMBINANT 19086313 h influenzae type b 19112612 Influenza Virus Trivalent-Split 2004 19112614 Influenza Virus Vaccine TV P-SURF 2004 19112615 Influenza Vaccine Trivalent 2004 (Live) 19121608 Influenza virus vaccine 43532251 antithrombin alfa 43533108 Influenza Virus Vaccine, Live Attenuated, A-Texas-50-2012 (H3N2) strain 46233853 oldaterol

APPENDIX A
Finding Ingredients that used to map to themselves in JULY-176-2015 but do not in MARCH-11-2016.

WITH CTE_OLD AS (
	SELECT c.*
	FROM [VOCABULARY_v5.0_20150717].dbo.CONCEPT c
		JOIN [VOCABULARY_v5.0_20150717].dbo.CONCEPT_ANCESTOR cr
			ON cr.ANCESTOR_CONCEPT_ID = c.CONCEPT_ID
		JOIN [VOCABULARY_v5.0_20150717].dbo.CONCEPT c2
			ON c2.CONCEPT_ID = cr.DESCENDANT_CONCEPT_ID
	WHERE c2.DOMAIN_ID = 'DRUG'
	AND c.DOMAIN_ID = 'DRUG'
	AND c.VOCABULARY_ID = 'RxNorm'
	AND c2.vocabulary_id = 'RxNorm'
	AND c2.CONCEPT_CLASS_ID = 'Ingredient'
	AND c.CONCEPT_CLASS_ID = 'Ingredient'
	AND c.CONCEPT_ID = c2.CONCEPT_ID
), 
CTE_NEW AS (
	SELECT c.*
	FROM [VOCABULARY_v5.0_20160311].dbo.CONCEPT c
		JOIN [VOCABULARY_v5.0_20160311].dbo.CONCEPT_ANCESTOR cr
			ON cr.ANCESTOR_CONCEPT_ID = c.CONCEPT_ID
		JOIN [VOCABULARY_v5.0_20160311].dbo.CONCEPT c2
			ON c2.CONCEPT_ID = cr.DESCENDANT_CONCEPT_ID
	WHERE c2.DOMAIN_ID = 'DRUG'
	AND c.DOMAIN_ID = 'DRUG'
	AND c.VOCABULARY_ID = 'RxNorm'
	AND c2.vocabulary_id = 'RxNorm'
	AND c2.CONCEPT_CLASS_ID = 'Ingredient'
	AND c.CONCEPT_CLASS_ID = 'Ingredient'
	AND c.CONCEPT_ID = c2.CONCEPT_ID
)
SELECT 'OLD ONLY' AS TYPE, * FROM ( SELECT * FROM CTE_OLD EXCEPT SELECT * FROM CTE_NEW) z
UNION
SELECT 'NEW ONLY' AS TYPE, * FROM ( SELECT * FROM CTE_NEW EXCEPT SELECT * FROM CTE_OLD) z
ORDER BY 1,2
#2

@ericaVoss:

The examples you listed are all deprecated, with the exception of 43532251 “antithrombin alfa”, which is a precise ingredient. Follow the “Maps to” or “Form of” relationships.

#3

Ha! Sorry should have seen this. Now I know the issue, there are some items in LAERTES being tagged with deprecated concepts.

#4

@Christian_Reich,

This topic came up in a conversation with @lee_evans and @anthonysena today. They asked why a RxNorm concept like this gets deprecated. My best answer was either some codes are replaced by other codes or some relationships to codes just retire. I realize this was not a detailed answer nor could I figure out what it was replaced with.

So the question specifically was why does something like 19121608-Influenza virus vaccine deprecated?

Additionally, it looks like some of the procedure codes still map to this deprecated concept as an ingredient.

SELECT *
FROM CONCEPT c
	JOIN CONCEPT_RELATIONSHIP cr
		ON cr.CONCEPT_ID_1 = c.CONCEPT_ID
	JOIN CONCEPT c2
		ON c2.CONCEPT_ID = cr.CONCEPT_ID_2
WHERE c.CONCEPT_CODE = '90658' /*CPT4 - "Influenza virus vaccine, trivalent, split virus, when administered to individuals 3 years of age and older, for intramuscular use"*/

I we reviewing this on v5.0 11-MAR-16

#5

@ericaVoss:

Because there is an Influenza Virus A and an Influenza Virus B. Ask the virusses why they can’t stay together as one. In addition, they are actually different each year:

select * from concept where lower(concept_name) like '%influenza%virus' 
and vocabulary_id='RxNorm' and concept_class_id='Ingredient' and standard_concept='S'

The mapping: We are still fixing those. It’s a crazy problem. Keep them coming.

#6

@Christian_Reich - Thank you, so in this case the 19121608-Influenza virus vaccine was retired for more specific codes.

#7

That’s what it looks like. They don’t comment on any of these choices.

#8

And that causes trouble because now the CPT code is no longer unambiguous.

t