Proposal Owner: Gowtham Rao

Use cases: http://forums.ohdsi.org/t/new-table-for-encounters-encounters-care-transitions/2297 http://forums.ohdsi.org/t/how-to-represent-encounter-payer-information-in-the-model/369 http://forums.ohdsi.org/t/capturing-more-than-one-care-site-id-or-provider-id-per-encounter-occurrence/1901 Also addresses using new Definition table the concern of atomicity and lineage of data to source http://forums.ohdsi.org/t/purpose-of-the-omop-cdm-analytics-ready-or-close-to-source/2358

See discussions: http://forums.ohdsi.org/t/definition-of-encounter-transfers-and-referrals-and-distinction-of-er-from-hospital-encounters/1311/4?u=gowtham_rao http://forums.ohdsi.org/t/new-table-for-encounters-encounters-care-transitions/2297

Proposal intent and problem statement:

• The intent of this proposal is to capture granular information in OMOP Common Data Model about encounters/microsivists/visits as is from the source data.
• Current Visit_occurrence table is a derived/calculated/inferred table - equivalent to “_ERA” tables. Current Visit_occurrence table does not meet use-cases that need the granular occurrence level information (sometime with lineage to source data).
• We propose visit_occurrence and a new visit_occurrence_era tables, where visit_occurrence will be repurposed to represent data as is, while a new table that is a replica of visit_occurrence table will become visit_occurrence_era and will be equivalent to current visit_occurrence table. (Current Visit_occurrence table becomes visit_occurrence_era and is equivalent to Drug_ERA or Dose_ERA.) Visit_occurrence is used to derive thru computation visit_occurrence_era table in a manner similar to the derived _ERA tables.
• Since we are proposing the visit_occurrence and visit_occurrence_era to have similar table structure, we will retain backward compatibility.
• Multiple visits must be related to each other. We can represent that in visit_occurrence and visit_occurrence_era using chaining/sequential method or parent-child/part-of.
• Example: an entire inpatient stay in one hospital may be one 'higher level' visit, with several smaller visits at ER, ICU, medical floor, rehabilitation floor etc. Each visit may have different start/end date-times. Example of sequential visits –> ER leading to ICU leading to medical floor, leading to rehabilitation. Example of parent-child visit: a visit for dialysis while in ICU.

Proposed visit_occurrence table: Will have the same structure as current Visit_occurrence table. We will have place_of_service_concept_id, parent_visit_occurrence_id and visit_occurrence_era_id

Relationship between visit_occurrence and clinical events tables like condition_occurrence, procedure_occurrence, specimen etc.

Use existing visit_occurrence_id (Fk) in these (event)_occurrence tables.

Representation of US claim data

US claims data generally has header/summary data and line/detail level data. Detail is a child of the summary. If parent_visit_occurrence_id = visit_occurrence_id then it is header/summary data. If parent_visit_occurrence_id is not equal to visit_occurrence_id then it is a detail. We create new visit_type_concept_id for the following

1. Facility claim (Summary)
2. Facility claim (Detail)
3. Professional claim (Summary)
4. Professional claim (Detail)
5. Professional claim (Any)
6. Facility claim (Any)

Proposed new visit_occurrence_era table: Will have the same structure as current Visit_occurrence table.

Linking visit_occurrence table to visit_occurrence_era will allow for provenance/lineage.

There may be n:m relationship between visit_occurrence and visit_occurrence_era, however one of them may be primary. This may be represented using primary_visit_ocurrence_era_id on visit_occurrence table.

All other non-primary relationship may be represented using FACT_RELATIONSHIP tables.

Relationship between visit_occurrence_era and clinical events tables like condition_occurrence, procedure_occurrence, specimen etc.

By joining thru visit_occurrence table

• Join visit_occurrence_era to visit_occurrence.
• Join visit_occurrence to (event)_occurrence tables.