OHDSI Home | Forums | Wiki | Github

Guidance needed: Representing scheduled activities in a prospective study

I am a working on a prospective study and we use OMOP for data representation for analysis. The study has the initial baseline and then follow-up data collection activities every 2 months (surveys, in-person clinic visits).
We need a way to represent these groups of fact records collected at each time point in the study. What would be the most appropriate approach?

  1. FACT_RELATIONSHIP linking all fact records to an artificially created OBSERVATION record that represents “data collection window”
  2. VISIT_OCCURRENCE for each data collection window and populating visit_occurrence_id in all fact records
  3. EPISODE for each data collection window and using EPISODE_EVENT to link fact records.

Thank you

Hello @opatterson,

Technically, all of the above could work.

If all follow-up collection activities happen during one visit, then you have the FK visit_occurrence_id in each clinical event table. Atlas also has a “occurs in the same visit” association when building your phenotype.

Fact Relationship can connect any two facts. It’s not easy to implement and not easy to use when you are connection > 2 facts. Atlas doesn’t use this field. Generally, this is not recommended for linking >2 events.

Episode can be to group events. Per [CDM v5.4 conventions,] (OMOP CDM v5.4) it looks like there should be a hyperlink to an article showing how to accomplish the link. But the hyperlink is missing. Atlas doesn’t use the Episode and Episode Event tables.

Thank you for your response.
My main struggle is that we do not have a data record for a “visit”. We only have two-week windows where multiple data points get collected through an online survey app or a clinic visit where a patient actually interacts with the study personnel but it is a part of the same data collection window.
Would it contradict the OMOP data conventions to create a visit record during source to OMOP ETL process that would group these fact records?

t