OHDSI Home | Forums | Wiki | Github

Measuring site information for blood pressure

Developers
#1

We are now collecting blood pressure (BP) information from our EHR data. It amounts about 16M measurements. Measuring site information such as central vein, Left/Right arm, Left/Right leg is important when we understand the clinical context of the BP measurements. While transforming the BP data into CDM 4.0 (and also into CDM 5.0), we found that the observation model in the CDM 4/5 does not consider this observation body site. Will this information can be included in the following new CDM model?
Rae

#2

Hi Rae,

You can already code this information using different LOINC concepts, for example:

Cheers,
Martijn

1 Like
#3

Hi, Martijn Schuemie

Thank you for the reply.
Do you mean that the LOINC concepts could be filled at the observation_concept_id column of OBSERVATION table?

For example,
– some BP record has ‘3001223 (iliac artery - left diastolic blood pressure)’ as observation_concept_id
– some BP record which dose not contain measurement site, has ‘45876174 (Blood pressure)’ as observation_concept_id

Best,
Eun kyoung Ahn.

#4

Hi Eun kyoung Ahn,

Actually, ‘45876174 (Blood pressure)’ is a classification concept and should not be used in the CDM. (You can see this in the new version of the vocabulary by the standard_concept field in the concept table). Only standard concepts should be used in the CDM.

Do you have measurements for which you only know they are blood pressures? Or do you know a bit more, which would allow you to select a standard concept?

Best regards,
Martijn

#5

Most of our BP data, we have only know they are blood pressure.

I found two concept_id at snomed code.
– ‘4152194 (Systolic blood pressure)’ and ‘4154790 (Diastolic blood pressure)’

Are they OK for the observation_concept_id of blood pressure?

Thanks,
Eun Kyoung Ahn

#6

@EunkyoungAhn:

Yes, it is. Use the following SNOMED concepts:

4152194 Systolic blood pressure
4154790 Diastolic blood pressure

Martijn correctly pointed a couple problem that we will have to fix:

  1. Should we “demote” less specific LOINC concepts so they can be used in the data?
  2. All vital signs in SNOMED should be domain Measurement, not Observation
  3. We have a duplication between SNOMED and LOINC

We will fix 2) 3) is work in progress, but it will take a while.

1 Like
#7

Thanks Martijn and Christian!
We will use the LOINC concepts that Martijn suggested for the BPs with specific sites.
All the remaining BPs, actually 13M out of 15M, will be designated with SNOMED concepts (4152194 Systolic blood pressure, 4154790 Diastolic blood pressure).
Best

t