OHDSI Home | Forums | Wiki | Github

CPT4 Vocab access and use

CDM Builders
,
#1

Hello,

I have been using OHDSI for about 6 months now but I was not involved at all in the creation of my instance. I need to make an upgrade of procedural codes using the CPT4 vocab I downloaded from ATHENA. From here I have a few questions I would appreciate assistance with. I would also appreciate any links to videos or sections of OHDSI github that may house more detail to the following.

  1. I don’t see a csv or xls that features a source_code_description. I also see I need an API Key to access data specifically for CPT4. Do I need to use the java package with an API key to finally see source code descriptions? My company does not have an account able to receive and API key, what do I do?

  2. What is the preferred format of using the vocabulary packages to map to existing tables something like fabric? Is this also meant to be done in Java, do I build out a code via python, or is there another option?

#2

Since I want to adjust a table housed in Fabric and not my SQL server, Do I still use Whiterabbit?

#3

@Clowson:

  1. You need to get the API key from the UMLS. It’s free, for anybody in the world. But that is the only way to get the CPT4 descriptions. We cannot distribute them according to the license. The CPT4 jar will do that for you automatically. No need to

  2. Not familiar with Fabric, but generally the use cases we use are mostly operating using SQL and R. Generic BI or other analytical tools don’t work well, because you need to apply logic that makes sense to healthcare. There is a lot of that existing in Github, I wouldn’t reinvent all those wheels.

The best way to get a good understanding is to read up in the Book of OHDSI.

#4

@Christian_Reich Thank you.
How are the _source_values added to the codes? I believe it to be during ETL, but where are the source values pulled from?

#5

Again, it’s all in the Book of OHDSI.

But briefly: You look them up from the codes in your source data. Let’s say you find a CPT4 code. You look it up in the vocabulary using Athena or SQL or any other method. The concept will have vocabulary_id=‘CPT4’ and concept_code=. There will be only one. That concept is the source concept. You insert a record with it in the source_concept_field of the table the domain_id tells you. Each table has one. Then you map it to the standard concept using the CONCEPT_RELATIONSHIP table, using the concept_id of the source concept and the relationship_id=‘Maps to’. The resulting concept_id_2 goes into the standard concept_id field of that table. Usually, that is only one record, but there could be more, which means you insert another record. You fill in the remaining fields and the record is ready for prime time.

t