OHDSI Home | Forums | Wiki | Github

Classifying parity of a pregnancy


#1

Hi,

I was wondering if anyone had any experience identifying the parity of a woman’s pregnancy event in EHR data? So many pregnancy-related outcomes are related to parity, but I don’t see an obvious place where it would fit into the OMOP model, with the exception of maybe text mining notes?

I appreciate any thoughts on the matter.


(Karthik) #2

Hi @scannemo, that is tricky. I know at our institution we worked with the OB team to standardize a note where the information was captured as a discrete field; otherwise, this is an information extraction task in the note. Currently, we don’t put the parity information in our OMOP, but I would imagine it belongs in the measurement table.


(Evan Sholle) #3

I agree with @cukarthik - this could go in the measurement table. Although I could also see an argument for putting it in OBSERVATION. However, we don’t capture it as a structured field in the EHR either. If your institution does capture it in that fashion, I’d recommend mapping it manually to a standard concept and populating MEASURMENT/OBSERVATION accordingly. If it doesn’t, you’d have to do NLP on the prenatal care progress note to get it out, then populate it either into NOTE_NLP or MEASUREMENT depending on how you handle the output of NLP.


(Jeremy Warner) #4

Hello, Adi Bejan here at Vanderbilt has done some work in this area; he is not on the forum but gave me permission to share his contact info. I’m also attaching an abstract that he presented at AMIA. As you can see, this was an NLP extraction task.

Bejan, Cosmin Adrian adi.bejan@vumc.org

2019_AMIA-IS_BejanDenny_ObstetricHistory.pdf (110.3 KB)


#5

Thank you for your thoughts. It’s a shame that it requires the infrastructure of NLP, but I greatly appreciate the link to Bejan’s talk. We are in the process of switching over to a new EHR system-do you think it’s feasible to catch someone’s ear at this point and have it be a measurement?


(Karthik) #6

It’s possible, but it depends what point you are in the switch over to the new EHR. In my experience, you need a clinical department or clinical leadership supporting the effort for it to happen.


#7

Appreciate the insight.


(Qi Yang) #8

@scannemo

but I don’t see an obvious place where it would fit into the OMOP model

Not sure if this helps. But there is a standard concept 4041279 - Parity finding. It belongs to Condition domain. There is an hierarchy under this concept, which includes (subsumes)

Nullparous (4012561)
Para1 (4030563)
Multiparous (4102166)
Gestation results in multiple offspring (42594973)

and Multiparous (4102166) also subsumes following concepts:

Para 2 (4167662)
Para 3 (4172590)
Para 4 (4173934)
Grand multipara (437937)

These are all standard concepts (SNOMED) that belong to Condition domain.

There is also similar concepts and hierarchy for Gravida - finding (4182766)


#9

thank you! It was very helpful to have that start as we went looking. It appears as if those codes are very infrequently used in our institution, but I’m glad to know where their home would be.


t