Mapping Transfusions and Donations to OMOP

CDM Builders
1

Hi all,

I’m currently working on mapping transfusion and blood donation events into the OMOP CDM, and I’ve run into a few challenges I’d love to get some advice on.

Specifically:

  1. Mapping Donations and Transfusions:
  • For blood donations, it doesn’t seem to be a rich set of standard procedure or observation codes in SNOMED or other vocabularies that clearly capture the act of donation (e.g., blood, plasma, platelets). I use generically the SNOMED code 4097300.
  • Transfusions are mapped in the device exposure table.
  1. Linking Donations to Devices:
  • I’m wondering how to approach this — especially using the device_exposure table in OMOP, and how/if links between person-level procedures (like donation or transfusion) and devices can be established meaningfully.

Given the limited standard codes in this domain, I’m curious:

  • If anyone have worked on modelling blood donations and transfusions?
  • Any advice on using custom concepts in this particular case?
  • Any best practices for representing the relationship between donation events and devices?

Any advice would be very much appreciated!

1 Like
2

Hi @anabarbararc:

  1. Yes, the donation is not a healthcare intervention. Well, it is, of course, but not an intervention to the patient. It’s like production of a medicinal product. We don’t consider that a healthcare event, and usually it is not recorded in any EHR or insurance claim database anyway. Are you interested in researching outcomes for the donors?

  2. The transfusion is a device exposure, because OMOP considers transfusions devices, rather than drugs. However, there has been debate about this in the past: Mapping Transfusion Data. But not matter where it goes, the person_id foreign key field should like any exposure (drug or device) to a unique patient identifier, so you can find out what else happened to the patient. Is that what you are trying to achieve?

3

Hi @Christian_Reich,

  1. I’m interested in mapping a blood donation observation, including information about (i) the methods of blood component processing and collection and (ii) the donor’s blood group information.
  2. The transfusion data is being mapped to the device exposure table as you mentioned. What I am trying to achieve is to link a donation to a blood product and consequently to a patient.
4

Understood. My question was about the analytical use case, not about the data engineering.

  1. The collection and processing (usually also storage) should be stored with the recipient, not with the donor. Use the ISBT vocabulary for that. It has all the different combinations or collection, processing and storage pre-coordinated.
  2. Unless you want to study outcomes for the donor (becoming a patient for some reason) all you need is the blood product units in the recipient data (where usually the need and the outcome arises). Do you?
5

Hi Christian,

  1. I appreciate the advice - I will have a look into it.
  2. I realise I may not have been clear about the the research project I’m working on. For the donors, we have demographics information as well as extended blood group data, which are relevant for the research outcome. I want to emphasise that donors are not patients in this context.