OHDSI Home | Forums | Wiki | Github

"Smoker" in the SNOMED vocabulary : Condition Occurence

We are coming across a few records related to “smoking, ex-smoker” which are mapped to Condition Occurrence due to the snomed vocabulary. Could this be something that gets changed in the next version of the vocabulary? The mapping type should not be condition, correct?


Is this the chance for us to create a convention for smoking and other lifestyle choices? I believe you essentially have the solution. It keeps popping up, now it happened at IMS, and I don’t have a good answer for Urvi. Let us know.


Yes, we can take conventions and vocabulary that we created for CDRN as a foundation. However, it will need a thorough review and input from practitioners and analysts. I will review it and post over the weekend.

In the mean time, wouldn’t make sense to change the domain id to ‘Observation’ for the SNOMED concepts ‘Never smoked tobacco’ (id = 4144272) and ‘Ex-smoker’ (id = 4310250)?


That was my initial reflex as well. However, “mean time” is not that quick with regard to making vocab releases, as you painfully know, and also we need to solve this properly. It is important.

Attached is a draft proposal for representation of tobacco smoking behavior and tobacco type in OMOP CDM v5 to support the PCORnet CDM requirements. Since these requirements are partially derived from the Meaningful Use Core Measures (also attached) and tobacco smoking representation in EMR, they should be a good starting point for comprehensive tobacco conventions for OMOP CDM. What’s lacking in this proposal is analysis of how the proposed concepts and conventions will support analytical queries.

I encourage all the previously contributing or interested members, @Christian_Reich, @DTorok, @toanong, @mkhayter, @schillil, @dckc , @ushah, @Vojtech_Huser, to get together and expediently turn these into a comprehensive proposal. Please let me know if interested.

Thank you.


Meaningful Use Core Measures

1 Like

That is a good start. I like the idea of specifying related facts and restricting value sets.

The current proposal is missing codes you suggest for tobacco smoking behavior and tobacco type.

CDEs (common data elements) try to define “standard forms”. See example here:

The tabaco type should include a “vaper” (e-cigaret user) as well. I wander how # of cigarettes translates to “vaping”

Thank you. I am currently enhancing one of our OMOP ETL processes–specifically its treatment of this area (i.e. smoking status and other Meaningful Use Core Measures)–and find your proposal to be quite useful and consistent with our developing approach.

Hi Rimma,

Any updates on this proposal being implemented ? We currently are working on a lot of data that would benefit from this.


Also wondering if there are any updates on this proposal. I very much like the draft proposal. I would add to the proposal to move concept_id=4052032, “Stopped smoking” from Condition domain to the Observation domain. And advocate for the “convention” to become that this concept is where smoking stop date is canonically recorded.

Will try to finish this in this cycle.

I logged this to an issue so we can keep track of it:

I have been working with a group on conventions on how to store information in OMOP CDM with the intent that it will be ETL’ed from OMOP to PCORnet. Here is how we plan to record smoking information. All the smoking related concepts are stored in the Observation table under a single observation concept id 4041306 (Tobacco use and exposure) and the specific smoking concept values are in the value_as_concept_id. This convention makes it easy to find all the related smoking information for a person.