Write down convention for patients "still in hospital"

[clairblacketer]

If patients are still in the hospital at the time of the ETL build, set the visit_end_date to the date of the data extract and set the discharge_to_concept_id to 32220.

[MelaniePhilofsky]

I'll write this up

[MelaniePhilofsky]

We updated the visit_occurrence.end_date, visit_occurrence.end_datetime, visit_detail.visit_detail_end_date, and visit_detail.visit_detail_end_datetime User Guide sections to direct users to set the visit_end_datetime to the date of the data extract when an inpatient visit is ongoing and the *.discharged_to_concept_id = 0.

We updated the visit_occurrence.discharged_to_concept_id and visit_detail.discharged_to_concept_id ETL Conventions sections to notify users it is assumed a Person returns home after a Visit, therefore, there is not a standard concept_id to represent 'home' and set the discharged_to_concept_id = 0

[MaximMoinat]

We should update this user guidance, as some things have changed.

Concept 32220 - Still Patient is non-standard, and the advice in the user guide for visit_end_date is to add this as the visit_type_concept_id, not discharge_concept_id.

(...) For Inpatient Visits ongoing at the date of ETL, put date of processing the data into visit_end_datetime and visit_type_concept_id with 32220 “Still patient” to identify the visit as incomplete. (...)

This is now conflicting with our conventions for type concepts to be standard and of class 'Type Concept'. Do you know what changed @MelaniePhilofsky @clairblacketer?

[cgreich]

Concept 32220 - Still Patient is non-standard.

Needs an issue for the vocabs: https://github.com/OHDSI/Vocabulary-v5.0/issues

[clairblacketer]

This has been open on the vocab github for a while: https://github.com/OHDSI/Vocabulary-v5.0/issues/505

[clarkevans]

Perhaps this is a data model issue, and in future CDM version, visit end date should be optional. Existing queries would ignore these visits, which is probably OK if they are not improved.

This is a problem for neonatal studies or those studies with long hospitalizations that would miss critically ill among the population. I think clever use of terminology doesn't make handling this case obvious. query logic should handle it more explicitly.

[clairblacketer]

Transferred from the CDM repo, I believe this is a THEMIS issue and still needs consensus on how patients till in the hospital should be handled.