OHDSI Home | Forums | Wiki | Github

How to Build Cohort Definition for Telehealth Utilization?

Hi OHDSI Community!

Hope you are all doing great today!

Today, I would like to open a discussion on building definitions for telehealth visits. As a starting point, I created the following concept sets:

These are according to Centers for Medicare and Medicaid standards. Now, what I am unclear on is how best to take these concept sets and create definition(s) from them. My current attempt at a full definition is here:

Basically, my definition takes the previous concept sets, looks for anyone in the CDM matching to one of these concepts from the concept sets above within the observation and procedure fields (based on investigation that is where those come from), and uses any single visit as an entry criteria to say, “this person utilizes telehealth services.” to build my actual cohort. From there, then I’ll further examine this cohort.

My outstanding questions on this are:

  • Should I further add constraints around entry criteria for utilization? Such as multiple visits, more than 1, etc? I’ve been doing a literature review with my group around this and we haven’t seen this kind of question answered.
  • Although I am using CMS guidelines as a foundation, should I tweak the core concept sets at all or am I missing something? I know this information will most likely be updated given a new US bill being passed extending telehealth coverage/services up through 2024.

Could anyone help me with this? Any workgroups I should join to maybe pitch this question to?


~ tcp :deciduous_tree:

@TheCedarPrince , I don’t know much about the specific telehealth concepts you’ve picked, but just in terms of creatng a cohort to return back all events of telehealth, here’s a cohort revision based on your intial definition: ATLAS

The main difference here is to use ‘all events’ in the entry events, and then specify cohort exit logic that ends with each event event. Then, 2 telehealth codes on the same day will be collapsed, but two codes on differnt days will be returned as separate cohort eras.

1 Like

I have no stake in the specifics but the classes seem awkward from a long-term perspective. In five years when COVID is just another endemic disease those names are going to feel pretty strange. “COVID” is the exception, not the norm. Therefore I posit the names should be:

1 Like

You are missing this Telehealth visit.

1 Like

Thank you. I was only concerned about the two awkward ones but should have continued with a complete list. But now that you mention it, this might be an even better form. Think about searching for these concepts in a list. This format groups them alphabetically.

1 Like

Hi folks - sorry for the delay in response! Thank you all so much for the help!

@Patrick_Ryan - thanks for the help in cleaning up the logic I had.

@MPhilofsky - thanks for the catch, will add that code in.

@jmethot - thanks for clarifying the names of the sets.

1 Like