OHDSI Home | Forums | Wiki | Github

Adding additional TYPE concepts

The blood data with which I am dealing today comes to us from EHRs and Transfusion Services tracking systems. To note the provenance of records being loaded to Device_Exposure and Observation tables, I expect to use device_type_concept_ID and observation_type_concept_id, respectively.

While there are several TYPE concepts that can be used for EHRs, I do not see one for Transfusion Services.

The closest that I see are from other domains:

id code name class concept valid domain vocab
45761156 467148005 Blood bank information system Physical Object Standard Valid Device SNOMED
45760595 466375004 Blood bank information system application software Physical Object Standard Valid Device SNOMED
45769521 708194008 Blood bank service Qualifier Value Standard Valid Observation SNOMED
4288740 68938006 Blood bank inventory control Procedure Standard Valid Observation SNOMED

If I am not overlooking something, how may I request that additional TYPE concepts be added to the standard?



Those concepts are Device and Observation concepts. We currently don’t have a “Bloodbank” type concept. We just have EHR. Do you think the distinction is important?

I am currently working on a project that utilizes data from both EHR and blood bank/transfusion services records. As records from disparate sources are brought together in a single research database, I think it appropriate that those who ultimately make use of the data be able to differentiate provenance.

Actually, that’s not how it works. It’s the analytical use case that drives standard concepts. Not availability of data. We have enough attic cases in the community that are never used for anything.

So, the use case must research transfusion and distinguish the provenance of the information. To the patient and the outcomes it is utterly irrelevant.

REDS is an international project producing the most detailed transfusion database in the world. It is currently in year 25 and is likely to continue for dozens of years.

Given the complexities involved in blood acquisition, processing, and distribution and their relationship to patient care, including the determinants for usage and handling in the clinical environment, I suggest that knowing where events and determinations occurred and where they were recorded is far from “irrelevant”.


Maybe irrelevant is the wrong word. I am not debating that “acquisition, processing and distribution” of transfusion couldn’t have an effect on health. But the Type concepts denote where a fact was recorded, not whether or not it is recorded. Changes to the CDM or vocabularies and their propagation are costly. We therefore are asking for examples of an analytical use case where that particular fact matters. For example, if you want to filter out all records not coming from the blood bank because you learned not to trust them, or the like.

Do you have that?