OHDSI Home | Forums | Wiki | Github

How to map drug frequency, contraindications, administration actions in OMOP

Implementers
#1

Hi, we have a use-case monitoring prescribing practices, including where there are contraindications.
We are mapping our source prescriptions/medication requests to OMOP but not sure which OMOP fields we should map the following to:

  • administration actions: e.g. given / withheld / unhold
  • frequency e.g. every 6/12/24 hours, 3 times a day, once a week
  • contraindication comments

Appreciate any insights you may have. Thanks

#2

Generally, OMOP CDM data capture what happened. It’s works with the Closed-World assumption, meaning, if there is a record it happened, and if there is no record it did not happen. So, “withheld” or “unhold” (you probably mean “on hold”) is not captured since the patient was not exposed.

That goes to the SIG field in DRUG_EXPOSURE.

Not sure what this is. The doctor comments on his own prescription by declaring a contraindication? Or your comment? If the latter - this is outside the OMOP CDM.

#3

Thanks @Christian_Reich
re: administrations actions, this is an action that happened and is recorded in the medication administration record (MAR). Patients are prescribed drugs, but the prescription may not be administered (withheld, on hold) due to a contraindication, often the nurse describes this in free text in the notes/comments.
For example, the doctor prescribes a drug, but the nurse does not administer it and records it as withheld due to a contraindication rule e.g. the platelet count is below a threshold. Later when the counts were above the threshold, the nurse changes the administration records to “unhold” or “given” and the drug is administered. They may comment the platelet count is now within the normal range for the drug to be taken.

#4

Yes, @rosies, I understand. All that happens. But the CDM captures what happens to the patient, not what happens in the heads of the providers coming to the right conclusion. A drug can only exert an effect, good or bad, if it is in the blood stream or on the surface of the human body. Otherwise it is a comparator at best. And that is what the dominant use cases are about.

Makes sense?

But to help you with your use case: You could try finding Observation concepts that serve your need and come to Themis to give them the blessing of a convention. Let us know what you find.

#5

Hello @rosies,

Another suggestion would be to take your ordered medication data and compare them to your administered medication data for a person. Give the orders type_concept_id = 32833 and the administered drugs type_concept_id = 32818. Put all comments and notes into the Note table.

#6

FYI, this does not work for outpatient unless one has all pharmacies, in area, with some type of data sharing plan/method. We tried this for a different grant and it was a complete bust and said requirement was dropped.

#7

Thanks @Christian_Reich @MPhilofsky @Mark

We went with mapping the MAR to Observations.
We used concept_ids e.g. 4082383 Medication given, 4162734 Medication not administered, type_concept_id: 44788358 Medication Record (closest to Medication Administration Record)

The clinicians found it more intuitive to find records for drugs that had been prescribed in Drug_Occurrence, then select for Observations where Medication was not administered (and Measurements/Observations that reflected contraindications). They didn’t use the contraindications in notes in the end as it was a bit challenging to define rules for free text.

t