OHDSI Home | Forums | Wiki | Github

How to capture Patient History in CDMv4

CDM Builders
#1

Diagnostic codes for items such as ‘Family History of’, do not belong in the Condition table. Meaning the only alternative is the Observation table. However, I do not see a sufficient number of Observation Types to cover all the possible (or even a majority of the the frequently used qualifiers) diagnostic qualifiers. For example in the source data I am working there are the following:
PROBLEM LIST/PAST MEDICAL
HISTORY OF
FAMILY HISTORY OF
OTHER PAST HISTORY
PAST SURGICAL
PREGNANCY/BIRTH
MATERNAL HISTORY OF
PATERNAL HISTORY OF
DIAGNOSTIC STUDIES
FRATERNAL HISTORY OF
SORORAL HISTORY OF

SNOMED has a number of ‘HISTORY OF…’ or ‘FAMILY HISTORY OF…’ concepts, but I do not see generic concepts for ‘HISTORY OF’ or ‘FAMILY HISTORY OF’. Is there any solution other than to request additional qualifying concepts be added to the vocabulary for Observation Type?

#2

@DTorok:

You are totally right. In V4 you can’t really solve this problem well. The current mapping will map history of something to medical history. This has been solved in V5. There, you have a ‘Maps to’ link to “Medical History” and a “Maps to value” to the SNOMED concept representing the condition.

If you have to, you could use the vocabulary 5 to do the trick in V4 CDM.

#3

There’s still a lot of ambiguity around this, so I suggest we write up a
definitive consensus best practice around this.

My opinion:

We should be storing ‘history’, ‘family history’ and other qualifying
observations in the OBSERVATION table. We should make the qualifier the
OBSERVATION_CONCEPT_ID, and the value is the concept of interest. So, if a
person has family history of breast cancer, the OBSERVATION_CONCEPT_ID
would be ‘family history’ and the VALUE_AS_CONCEPT_ID would be the
corresponding standard concept for ‘breast cancer’. The
OBSERVATION_TYPE_CONCEPT_ID should not be used for the qualifier, that is
intended to be the provenence of the data, so did it come from EHR intake,
a claims record, etc.

In the CDMv5 specification (
http://www.ohdsi.org/web/wiki/doku.php?id=documentation:cdm:observation,
http://www.ohdsi.org/web/wiki/doku.php?id=documentation:cdm:condition_occurrence),
we make the ‘history’/‘family history’ use case explicit to be placed in
the OBSERVATION table, but I think we should do the same for other
qualifiers that people intend to use in their analytical use cases, and
reach consensus around the OBSERVATION_CONCEPT_ID to use for each such use
case.

#4

Let me write up what we have.

t