OHDSI Home | Forums | Wiki | Github

Code blue mapping

Hello everyone,

I’ven working on mapping clinical data to OMOP CDM for one hundred patients who incurred a code blue event while hospitalized.

My doubts are the following:

  • Should the code blue event be considered as a record in the PROCEDURE_OCCURRENCE table?
  • Do you know what code you should map a code blue event to (e.g. CPT-4)?

Our purpose is to record the code blue event, outcome and date of occurrence. Therefore, I would like to know what is the best practice to do it in OMOP CDM.

Regards,
Alonso

Hello,

What do you mean by code blue?
It can be any type of cardiac/respiratory arrest, right? But since the patient is in hospital, CPR is often applied and more data is generated on the way, like the reason for code blue: ASY or VF, etc. Drugs are given, procedures take place, etc.

If you know nothing but code blue, you may go with mapping to Cardiorespiratory arrest.

Mapping to procedures and other domains depends on the granularity of your data.
CPR can be mapped to Cardiopulmonary resuscitation

And outcomes can be analyzed as outcomes of any other event with help of OHDSI analytical tools.

Sounds good?

Hello @zhuk, by code blue I mean the following. Events related to these situations:

  1. Commitment of conscience
  2. Overt respiratory distress or cyanosis
  3. Trauma/fall associated with severe functional impotence
  4. Seizure
  5. Massive or exsanguinating hemorrhage

However, in our database we only know the date of the code blue, but not the situation that caused it. At best, we have access to its outcome: death, transfer to a critical patient unit, or false alarm.

Regards,
Alonso

That’s a lot of conditions inside one definition.

If you can’t find any suitable standard concept in Athena either (I tried to find something, but definition is too broad and no concept fitted well), the good way to unite code blue events is to create your own custom concept. You may even make it standard if you don’t plan any network research and you are the sole owner of the vocabulary.

To create your own concept, you need to create a record inside the concept table, put it’s relationships in concept_relationship table like for any other concept. Just make sure that concept_id is more than 2 bil.

@Alonso:

There is no such a thing as a code blue diagnosis or procedure. Looks like you have to make a list of all the conditions. Whether or not they actually led to a code blue alarm in the hospital - anybody’s guess.

Thanks @zhuk and @Christian_Reich four your suggestions. I was analyzing the problem better and arrived at the following (Details in the image below):

  1. The Code Blue will be mapped to the concept “Signposting to rapid response team” considering it as a procedure.
  2. The outcomes derived from the Code Blue procedure will be treated as observations, such as: Death, Patient transfer, and Stabilized.

What do you think about this approach?

It depends on your CDM version. On version CDM 5.3 and 5.4 death table is used to store all the death related data, so use it. On version 6.0 death table is removed and fields to person table added, so you have to use them. Check the specifications on different versions here.

However, you can still map death to condition/observation ‘death’ and use these tables instead of specific one, but this approach is strongly not recommended.

Patient transfer to intensive care unit - use visit tables

Stabilized - I think, the better shot would be Observation (observation_concept_id - patient status, value_as_concept_id - stabilized) or even Condition, but not a standalone Observation with concept class id of Qualifier value

Signposting to rapid response team - great as a procedure

That’s only one of the possible approaches to use, yours would be fine too in analytical perspectives, but it’s not canonical and if you would ever want to collaborate with other researchers, you won’t be able to, until they have the same data structure. Also, your mapping will determine tables where to put records.

Good luck with this!

Hello @zhuk,
Thanking you for your response, I would like to mention that I am currently using CDM version 5.4.

As my intention is to collaborate in the future with other researchers, I would like to be canonical regarding the modeling of my use case in the CDM.

In the diagram below I have followed the flow of a patient in what we call “Code Blue”, where I have drawn related tables in which I must store the information within the CDM according to your suggestions.

Due to my background is in computer science, and also new to using OMOP CDM, I have the following doubt: There is no canonically correct way to treat death, transferred and stabilized as observations/ outcomes of the RRT team procedure? Is the canonically correct way to store the information as illustrated in the diagram?

Regards,
Alonso

Sorry for the late respond

Your diagram looks good to me

t