OHDSI Home | Forums | Wiki | Github

There are not enough Observation_concept_id in the OMOP CDM


(Wang Man) #1

Hi,guys! I’ve got a problem,please give me a hand.Thank you at first.
We’ve got some datas about “Pattern of menstrual cycle”.They are from “Patient report”.So we want to put them into the Observation table.But I found that “Regularity of menstrual cycle(4265552)” has no child concept.We can not record futher datas into the OMOP.
However,we found that “Finding of pattern of menstrual cycle(4095940)” whose domain is condition has enouge concept id.We can record all of our datas into the OMOP through Condition table.
So,what should I do now?Could OMOP CDM add more Observation_concept_id?


(Christian Reich) #2

@Vines:

You want to convert what apparently is an EAV in your data: “Pattern of menstrual cycle” - “irregular” and “Pattern of menstrual cycle” - “short” etc. But these things indeed are Conditions, but they should be pre-coordinated. So, you want things like 196168 Irregular periods and 4128325 Short menstrual cycle. So, take your EAV and find the right Condition Concept from http://athena.ohdsi.org/search-terms/terms?domain=Condition&standardConcept=Standard&page=1&pageSize=50&query=menstrual.


(Wang Man) #3

@ Christian_Reich:
You are a star!But I am still a little bit confused.
1.I’m sorry,what is EAV?
2.What you mean is to find a coordinated id in the Condition domain.But in my opinion,every domain of the concept id should be Observation if I want to record them into observation table.They are actually from patient report.I don’t think they should go to the Condition table.So I am not sure what you want me to do.
3.Can I use the event id and event concept id to record this kind of datas?


(Christian Reich) #4

@Vines

Just doing OHDSI community work. :slight_smile:

  1. Entity-Attribute-Value. In your case E-“Observation” - A-“Pattern of menstrual cycle” - V-“irregular”.

  2. Everything is an Observation, unless it is a Condition, Procedure, Drug, Device, Specimen or Visit. So, a irregular menstrual pattern is a Condition. Something wrong with the mental or physiological function of the organism. Observations are EAV, so you can split it into a question-answer pair. All the others are fully normalized relational tables, and there you have one concept to capture the content of the record. In your case that would be “Irregular pattern of menstrual cycle”, or, because that doesn’t exist just like that, SNOMED concept 196168 Irregular periods.

The reason we have an EAV model in OBSERVATION and MEASUREMENT is because it is impossible to pre-coordinate all possible Concepts. For example, a measurement could have 100 mg/dL, 100.0001 mg/dL, 100.00002 mg/dL, 100.00003 mg/dL, etc. An infinite number of values. So, we have to split it into an attribute (glucose concentration in blood) and value/unit.

  1. Event ID and event concept ID is for crosslinking records from different tables. So, the you want to say that a Condition was diagnosed by a certain Provider. So you cross-link the Condition record to a Provider record.

t