OHDSI Home | Forums | Wiki | Github

Representing Severity in OMOP


(Mark Wlodarski) #1

I was wondering about how other sites have handled implementing severity of an condition.

I am looking to represent the BOCK scale for food allergy reaction severity. This is a four point scale running from 0-3 where 0 represents a reaction not being present and 1-3 represent low, moderate, severe. 1-3 represent specific defined observable attributes, but those would be something understood by clinicians and researchers accessing the data, all I should need to do is indicate the level of severity.

I was wondering if anyone else has encountered a need to indicate severity, and how it was handled?

Thank you,
Mark


(Christian Reich) #2

@wlodarmt:

We don’t have the BOCK scale yet. We could introduce it, but we would need a use case. Also, we would need to make sure the data exists. Generally, clinicians record the actual skin, throat or systemic reactions, instead of using a scale they are not familiar with. Let us know.


(Mark Wlodarski) #3

Thanks @Christian_Reich
We would ideally be recording the items as an observation of the reaction in question. The BOCK scale is a symptom severity scale used for scoring the severity of food reactions during a food allergy test, such as an oral food challenge or skin prick test. This is something our Food Allergy experts have identified as something they would like to be included.

I have attached an example study that used the BOCK criteria for oral food challenges.
2012_SAMPSON_JACI.pdf (362.6 KB)


(Oleg Zhuk) #4

Hi, @wlodarmt

If you ask about severity representation in general, not only about Bock scale for food allergy, there is a way to handle it. The way, called fact_relationship table

I would suggest creating a combination of records in CDM tables: condition for food allergy and observation for severity of this allergy. You can create any other combinations, also including the type of allergy or specific conditions if you need it.

The overall result can look something like:

  1. Contition_occurence: event_concept_id - 4304894 Allergic reaction to food
  2. Observation: event_concept_id - 4037113 Symptom severity or find better one
  3. Observation: value_as_concept_id (low, moderate or severe) OR value_as_number (0 - 3)
  4. Fact_relationship: relationship_concept_id can be something like 4165382 Associated with

The other simple way to do this thing is to create one record in the Observation table with 4188027 Allergy to food as event_concept_id + concepts like 1177222 Low severity, etc.
You may lose mobility going this way, but that should work.


(Emma Vos) #5

Hi @wlodarmt,

We have been discussing this topic on capturing severity in the Clinical Trial Working Group as well. These discussions have led to the following two proposed conventions for linking an observation or measurement to another record, using attributes from OMOP CDM v6 and the oncology extensions :

  • Observation (v6 attributes)
    • observation_event_id
    • obs_event_field_concept_id
  • Measurement (oncology extension)
    • modifier_of_event_id
    • modifier_of_field_concept_id

The _event_id contains the primary key of the record that is to be modified (e.g. a condition_occurrence_id).

The _field_concept_id is a concept specifying to which table the given event_id points. (e.g. concept_id 1147663 refers to the condition_occurrence_id).

Using the first proposed convention (observation_event_id, obs_event_field_concept_id), you would follow steps 1-3 suggested by @zhuk. But instead of using the fact_relationship table, you would add 1147663 (condition_occurrence.condition_occurrence_id) as obs_event_field_concept_id and the condition occurrence you are referring to as the observation_event_id .


(Christian Reich) #6

@EmmaVos:

Is this submitted as an CDM change proposal?


(Leilei Zhu (Clinical Data Standards Lead, UCLH, UK)) #7

I think as an international community, we really need to work towards embracing or at least having an open mind about international standards rather than creating anything new.
The best way to standardise severity is using FHIR standard to support interoperability, the ultimate aim of having EHR systems.

FHIR has clearly defined the severity score: (noticed can’t capture screenshots here).

it is supposed to be a table view below.

Code Display Definition
mild Mild Causes mild physiological effects.
moderate Moderate Causes moderate physiological effects.
severe Severe Causes severe physiological effects.

https://hl7.org/FHIR/valueset-reaction-event-severity.html


(Emma Vos) #8

@Christian_Reich, not yet. We are currently working on a larger proposal and intend to share it in the upcoming week(s). And would say it is more of a CDM conventions proposal rather than a change proposal.


(Leilei Zhu (Clinical Data Standards Lead, UCLH, UK)) #9

Forgot to mention FHIR also defines the codeset for any of their “resources” i.e. severity. If you follow the link I sent, you will see.


(Mark Wlodarski) #10

Thank you all for your help! I definitely have a number of possible solutions within the existing OMOP framework. @LeileifromUK your reaction severity suggestion seems like it will fit the best.


t