OHDSI Home | Forums | Wiki | Github

How to populate transfusion rates

CDM Builders
#1

Hi @MPhilofsky @clairblacketer @Christian_Reich @karthik

CUIMC is attempting to conduct a study centered around IV fluids for correction of hypernatremia. For this, we need to ingest the transfusion rate values for already existing drug exposure records that indicate administration of continuous fluids. We internally thought the SIG field is the most appropriate column to place these transfusion rates. But we wanted to post this question on the forums to get the judgments of the community (and to learn if there is a convention for this). Thanks!

#2

Hello @thomasfalconer,

Interesting study!

You can utilize the Drug Exposure table as it is now to determine infusion rate. Subtract the drug_exposure_start_datetime from the drug_exposure_end_date time. Then use the result to divide it by the quantity. This will give you the hourly rate for the active ingredient.

The Health Systems interest group and others ETLing their source drug infusion data to the OMOP CDM would really appreciate it if you shared your ETL code from the source data to the OMOP CDM. We struggle with determining the best way to represent an iv drug infusion record with the variety of start, stop, pause, cancel, etc. flags along with different orders for the same drug which also start, stop, cancel, discontinue, etc. Also, I’m interested in how you determine an end_datetime when it isn’t specifically given in the source data. It’s quite the tangle!

#3

Infusion usually involves a device. I wonder if you are populating the device exposure table. If so, have you seen any issues?

#4

HI @MPhilofsky , thanks for responding. We don’t need to infer rate. We have rate (intended mL per hour) in our source data. We are wondering if there is a convention for how to populate it in drug exposure? We are thinking the SIG column is the best column for such information but wanted to check with others first. What is your judgment regarding SIG column versus QUANTITY column for this context?

#5

There is not a designated field for these data. If a variable / data point can be inferred from data in the CDM, we don’t add an additional field. Reason being, the data will conflict at times and then the analyst has to chose which is more true: The inferred data or the data in a specific field? The OMOP CDM is analyst ready, end users shouldn’t be making these choices because they are the ones furtherest from the source data. These decisions, called business rules, need to be made at the time of ETL.

t