Mapping ICD-9, ICD-10 and potentially MedDRA into a common dictionary

Implementers
1

Dears,

I need to perform signal detection activities on MarketScan data and potentially link the quantitative signals to whatever has been reported in our spontaneous report system.

MarketScan gets the diagnoses in ICD-9 or ICD-10 dictionaries whereas our spontaneous report system uses MedDRA 20.1 dictionary.

What mapping tools/data would you advise me in order to create a common framework? I know the question is vague and even keywords, general directions would help me.

Thanks,

Lionel

2

@lionelVH:

You are right there already. If you convert the data to the OMOP CDM and map all the ICD-9-CM and ICD-10-CM codes (not ICD-9 and ICD-10 as folks usually abbreviate them, those are WHO codes), your data will be coded in SNOMED, and MedDRA are hierarchical classifications of those. So, you can use MedDRA to define all your cohorts, and it will work.

If you need help with figuring out how the vocabularies work please go through the Tutorial.

3

Hello Lionel, was just about to answer the same as Christian. Hereby a random example I had already written down:

The SNOMED concept for ‘Essential Hypertension’ has mappings from ICD10 code ‘I10’, ICD9 code ‘401’ and the MedDRA code for EH (10015488). See http://athena.ohdsi.org/search-terms/terms/320128/table.

4

Hi @lionelVH,

@Christian_Reich and @MaximMoinat are right.
But the thing is that due to hierarchy inconsistency you may miss some information.
Can you please the post the list of the concepts of interest so we can check them

5

Hi,

Thanks all for your input.

@Dymshyts As the project is for signal detection I am not interested in a particular medical condition (or concept) as the project is hypothesis-free and will scan all diagnoses and spot those showing an unexpected pattern.

@Christian_Reich Thanks for the clarification; I registered to Athena to get dictionaries for ICD-9-CM & ICD-10-CM. I also requested the MedDRA one but it seems there is an extra step for this one.

6

You did, and yes, you have to obtain a license. We can’t issue licenses, we can only re-distribute to folks who have one. Here is how it works.

7

@Christian_Reich I got the files extracted for ICD-9-CM; I had a look into it and concept.csv seems to be the most relevant one to be used. However, even if they are supposed so be .csv files they don’t really display obvious separator. Are there OHDSI tools available to transform this flat file into a structured table? If not, what is the separator used in the csv file to separate fields?

EDIT: should I follow these steps: https://github.com/OHDSI/Vocabulary-v5.0/blob/master/ICD9CM/readme.md ?

8

@lionelVH:

It’s all here. Pick a database technology, and the first folder VocabImport contains the loading scripts.