Hi all!
I would like to share the solution which we implemented in our project. It is very similar to the one posted by
@nzvyagina with the only difference that we use modifier_of_event_id and modifier_of_field_concept_id fields proposed by oncology WG https://github.com/OHDSI/OncologyWG/wiki/MEASUREMENT to link specimen and the associated measurement records.
Source:
Specimen: Bld CVC
Culture: Bacterial blood culture
Organism identified in culture: Coagulase negative staphylococcus
Drug susceptibility test: 28-1 (‘Ampicillin [Susceptibility] by Minimum inhibitory concentration (MIC)’)
Drug result: >8
Susceptibility: Resistant
Serological test: 20966-8 (‘Staphylococcus sp identified in Unspecified specimen by Organism specific culture’)
Serological test result: Detected
CDM:
1st record: Specimen table.
Specimen_concept_id = 4045667 (Venous blood specimen)
2nd record: Measurement table. Store bacterial culture and organisms identified in specimen.
Measurement_concept_id = 3023368 (‘Bacteria identified in Blood by Culture’)
Value_as_concept_id: 36309331 (‘Coagulase-negative staphylococci’)
Modifier_of_event_id = specimen_id from the 1st record
Modifier_of_field_concept_id = concept_id with the name ‘specimen.specimen_id’
3rd record: Measurement table. Store drug susceptibility test.
Measurement_concept_id = 28-1 (‘Ampicillin [Susceptibility] by Minimum inhibitory concentration (MIC)’)
Value_as_number = 8
Operator_concept_id = 4172704 (‘>’)
Value_as_concept_id = 45878594 (‘Resistant’)
Modifier_of_event_id = measurement_id from the 2nd record to link drug susceptibility test with the associated bacteria
Modifier_of_field_concept_id = concept_id with the name ‘measurement.measurement_id’
4th record: Measurement table. Store serological test.
Measurement_concept_id = 20966-8 (‘Staphylococcus sp identified in Unspecified specimen by Organism specific culture’)
Value_as_concept_id = 45877985 (‘Detected’)
Modifier_of_event_id = specimen_id from the 1st record
Modifier_of_field_concept_id = concept_id with the name ‘specimen.specimen_id
To summarize, we create relations:
- Bacterial culture to Specimen
- Drug susceptibility test to Bacterial Culture
- Serologial test to Specimen
The problems which we encountered during the implementation and which could be discussed within community:
- Measurement.value_as_concept_id should belong to domain_id = ‘Meas Value’. Unfortunately, not all culture results belong to this domain. In priority, we mapped to ‘Meas Value’ and when it was not possible - then we mapped to ‘Observation’ and stored it to measurement.value_as_concept_id.
- Would it be convenient for everyone to use modifier_of_event_id instead of Fact_relationship table?
As this topic was discussed many times with different proposals, I believe we have enough examples and use cases to come up with the common solution which can be published in CDM conventions. @Christian_Reich, @SCYou, @Dymshyts, @rimma, @DTorok, @cukarthik, @mgkahn, @MPhilofsky, @philipzach, @mgurley, what do you think guys?