Thank you @DaveraG for the info and links! And thank you @mellybelly for offering to help! Visits in US EHR data are very challenging. The Health Systems Interest Group would be very interested in knowing how N3C creates a Visit Occurrence record and a Visit Detail record from source EHR data. Do you have an algorithm for identifying those pesky “administrative/transactional” encounters? The ones which look like a visit, but are actually artifacts of the EHR trying to keep PK-FK integrity for every clinical event entered into the EHR. The HSIG has been struggling with this issue for years!
The ambiguity of EHR encounters is different from the ambiguity of OMOP visits I describe in the first post of this thread. My concern with standard visit concepts is not only do they identify a visit (IP, OP, ER, etc.), but some also have Care Site and/or clinical event information. And some standard visit concepts are children to both an inpatient and an outpatient visit. It’s quite confusing for our analysts and researchers trying to create cohorts!
Things we need to do:
- Bring your use case! Historically, we used standard visit concepts as a proxy to severity. I’m only trusting a diagnosis of myocardial infarction if the clinical record is associated with an IP or ED visit.
- Let’s define “visit”. What does that mean from the patient perspective? For the most part, I thought we answered that question really well in our CDM v5.4 specifications. IMO, we need to get rid of “pharmacy, lab and ambulance” visits since those are care sites. And if the community decides to keep the combo Visit/Care Site, then this list needs to be greatly expanded. But if we do that, then do we get rid of the Care Site table? Researchers use the Care Site table to identify persons in the ICU when linked to a Visit Detail record. As always, there are pros and cons of every decisions.
I think we can accomplish #1 and #2 at the Symposium. If you can’t attend, please post here! After we complete the first two steps, we can come up with some proposals, create an issue with the Themis, CDM and/or Vocabulary WGs, and disambiguate the standard visit concepts.
@Alexdavv - we look to you for further guidance on the Vocabularyathon prep work ![]()