If there is data that starts from the discharge date, should it be recorded in the Visit Occurrence?

CDM Builders
1

Hi all,
I always refer to everyone’s posts.
Today, I would like to hear your opinions on how to handle discharge dates. Due to the method of extracting source data, we have cases where data starts with a discharge date without an admission date. In such cases, I would like to know your policy for ETL. I am considering three options:

  1. Generate records that only have a discharge date without an admission date.
  2. Do not generate records that start with a discharge date.
  3. Insert some value for the admission date and generate the record.

This is a very detailed matter, but I would appreciate hearing about your experiences.

2

#1 is against the rules and regulations of the OMOP CDM Visit Occurrence table. The start_date must be present.
#2 If you need these records, then they will be missing. Do you need these records? What are you asking of the Visit records? If you are looking at length of stay, then your data will be erroneous if you make up an admission date. In this case, I would say to remove the records. But it depends on your use case

1 Like
3

option 3. with imputed start date equal to known discharge date

(assuming it is inpatient visit).
I can imagine how there is some cut of date when the data era simply starts.
Patients in hospital on that data-era-start date can definitely exist. And start date prior data-era-start is logically “off limits”

1 Like
4

Thank you for your response.
I appreciate your opinions on the multiple options.
I also thought removal might be fine, but since there were hundreds of patients currently hospitalized, I am wondering if there is some impact, albeit small.
Considering the use case, I will discuss this with my colleagues.
Thank you!