All, I am looking for guidance on how to track procedure revisions or re-operations in the OMOP data model.When a patient has for example hip surgery this event is tracked in the PROCEDURE_OCCURANCE. In the event where there is an issue with the original procedure and either a revision or re-operation is required the revision or reoperation is again stored in the PROCEDURE_OCCURANCE table but I want to track the fact that this is a revision and link back to the original procedure. Is the recommended way to create this relationship to use the FACT_RELATIONSHIP table with a relationship of “revision of” or something link that?
Yes, this would be the right approach. According to OMOP conventions you will need to create two FACT_RELATIONSHIP records to indicate bi-directional relationships between surgery and re-surgery.
There is no designated relationship concepts for re-surgery. You may possibly use:
‘Occurs after (SNOMED)’ (44818783), ‘Occurs before (SNOMED)’ (44818881).
I’m not sure what kind of data you are working with, but if these surgeries are CPT/HCPCS codes, then there are modifiers that usually accompany these codes to indicate these surgeries are repeat/revision codes. Modifier 76 indicates that a repeat subsequent procedure was done by the same provider. You can store the modifier information in the modifier_concept_id and qualifier_source_value fields for the re-operation procedure_occurrence record.
This might be an alternative for you and you don’t have to worry about using the fact_relationship table. If you are not using CPT/HCPCS codes, then I believe the fact_relationship table is your best bet.
Apologies for the slow response, been travelling and frantically trying to catch up now that I am back in the office. Thanks @rimma for that feedback and the info on the SNOMED codes. I will investigate that further or possibly create my own codes to support this. @jenniferduryea, I done not have CPT/HCPCS codes in the data that I am loading so I will need to go with the fact_relationship option.
As a slight aside, do you use a similar method to store bi-lateral procedures? The procedure type for each joint can be different and so needs to be tracked separately but still is performed under a single procedure identifier. Issue with this is the procedure ID is no longer unique…