OHDSI Home | Forums | Wiki | Github

Plausible gender checks and ICD-9 > SNOMED mappings

(Evan Sholle) #1

Thanks to @Ajit_Londhe’s fix, I was able to run the DQ Dashboard on our CDM instance. One of the first things I noticed was that the plausibleGender checks flagged a pretty substantial error:

For a CONCEPT_ID 197605 (Inflammatory disorder of male genital organ), the number and percent of records associated with patients with an implausible gender (correct gender = Male). (Threshold=5%).

In investigating a bit further, I found that these were mostly source records with an ICD-9 code of 616.9 (Unspecified inflammatory disease of cervix, vagina, and vulva). In ETLing our old ICD-9 data into the CDM, we use the mappings in CONCEPT_RELATIONSHIP, which, it seems, is what threw this issue, as 616.9 maps to the SNOMED code referenced in the DQ check error message:

select * from FULL_OMOP.dbo.CONCEPT_RELATIONSHIP cr join FULL_OMOP.dbo.CONCEPT c on c.concept_id = cr.concept_id_1 where concept_id_2 = '197605'

@Dymshyts and other Athena folks, is this mapping functioning as intended? Any thoughts on how to preserve the gender-specific nature of these ICD-9 codes and avoid throwing DQ errors?

(Polina Talapova) #2

@esholle, thank you for reporting! To preserve the gender-specific nature of these ICD-9 codes, the vocabulary team has to fix mappings of the following ICD-9-CM concepts:
616.89 Other inflammatory disease of cervix, vagina and vulva
616.8 Other specified inflammatory diseases of cervix, vagina, and vulva
616.9 Unspecified inflammatory disease of cervix, vagina, and vulva
They will be re-mapped to the SNOMED concept of “40483613 442506007 Inflammatory disease of female genital structure”

Currently, we are working on the ICD-9-CM vocabulary review, so such changes will be released as soon as possible.

(Evan Sholle) #3

This is great, @Polina_Talapova - thanks!

I did notice that a few of the other women who had this concept had an ICD-9 code of 608.4 or 604.91, but I think those are just miscodings from the source data, as both of those ICD-9 codes seem specific to male genitourinary disorders.

(Patrick Ryan) #4

Thanks @esholle for finding and reporting this. I think you’ve discovered a new opportunity: using ‘data quality checks’ as a test for ‘vocabulary mapping tests’. I think we should look for more of these types of opportunities to continue to advance the quality of our vocabulary and data efforts. Great stuff!