OHDSI Home | Forums | Wiki | Github

How to map a country specific reimburesement codes to the CDM?

(Javier Gracia-Tabuenca) #1


I’m maping EHR from Finland to CDM. There is a code system specific from this country. Namely drug reimbursemen code.
When a patient expends a set amount of money in purchasing drugs they get a reimbuirsemen and this is marked in the EHR with a code indicating the type of disease/medicine group.

The coding has a two level herarchy, a piece of this coding is showed next

  • 103 - Diabetes, Insulin Treatment

    • 171 - Insulin degludec (limited special reimbursement)
    • 177 - Insulin glargine 300U / ml (special allowance limited)
    • 215 - Diabetes Other than Insulin Treatment
  • 137 - Uremia requiring dialysis

  • 123 - Chronic Vitamin D Disorders

This seems a very specific system.
I have two questions:

A) Is there an existing standard vocabulary to map this coding ??

B) If not, could I create a vocabulary the includes this codes and marke it as Standard, so that I can use it in my implementation of Atlast ? (I think atlast does not work if i try to use the non-standard codes)

C) Is there any other way ??


(Christian Reich) #2


It’s C). What you would want to do is to map these codes to SNOMED, and create CONDITION_OCCURRENCE records at the day of the dispensing.

That will capture the facts. It doesn’t capture the fact that the Drug and Condition are connected. In the OMOP CDM, we do not generally capture these cross-links for two reasons: (i) we very rarely get it, and when we get it the quality is low. For example, a diazepam can be prescribed against anxiety and insomnia. Or both. Now what? In most cases, it is not exactly clear if the anxiety causes the insomnia or the other way around. And (ii) because those connections are usually the outcomes of our use cases: What are the indication or consequences for Drug XYZ?

Let us know if you need further help with that.

(Javier Gracia-Tabuenca) #3

Thanks a lot @Christian_Reich.

Yes, capturing the connection to drug is now relevant. We also see it more as a indication of a condition.

But I was thinking if it would make more senses to clasify in the OBSERVATION table as a “medical history”, or if there is any other observation vocabulary that would fit this.

PS: Im just the data analist. Im trying to convice researchers and doctors in my organization to muse CDM and atlast. At some point some expert will handel this porperly, I’d exploring posibilities for when they ask.

(Christian Reich) #4

Not sure I understand. Indication is a condition.

We actually took that out. There used to be a relevant_condition_concept_id in the DRUG_EXPOSURE table, but it is never clean. The conditions are wishy washy, often there is not one but a few, etc.

What you want to tell them is that the indication or the outcome of a drug is something OHDSI research generates, rather than taking it as an input. So, the question is: All patients with Drug XYZ - what is the indication that doctors prescribe it for, and does it work? If you already know you don’t need to run analytics.

Makes sense?