parisni
commented
6 years ago on the way of doing so.
BTW, we added a specimen entry for each laboratory from labevents/chartevents. This allow keeping the charttime/storetime dates.
TODO are:
- mapping microbiology specimen to concept_id (manually -> 87 items)
- find/propose relevant relationship_id to define all those fact_relationship. Apparently existing relation are not focused on culture sensitivity -> should we propose some from snomed or loinc ?
aparrot89
We mostly agreed on how to store microbiology data, but there are some loose ends to tie up.
Specimen
Each microbiology culture should be associated with a specimen, and this specimen is intuitively stored in the SPECIMEN table. It's unclear how
disease_concept_idshould be used, so we have opted not to use it.Culture
In order to indicate that a culture was done, we should have a row in the MEASUREMENT table that indicates that a culture was performed. This row should have a relationship to the specimen using FACT_RELATIONSHIP.
Bacterial count
If one or more bacteria types were cultured, then for each bacteria the numeric measurement (e.g. colony count) should be stored in MEASUREMENT with the
value_as_numberbeing the count and theconcept_idbeing a standardized code for this count (I believe LOINC has a good vocabulary for this). Each bacterial growth should have a relationship to the specimen using FACT_RELATIONSHIP.Antibiotic
Tests for antibiotic resistance are stored in MEASUREMENT. At the very least, we will have one row per antibiotic tested which indicates whether the bacteria was susceptible, resistance, or indeterminate. Then, depending on the raw data, we may have another row which indicates the dilution of that antibiotic. This will link to the bacteria using FACT_RELATIONSHIP. Then, arguably, the dilution level of the antibiotic should link to the antibiotic susceptibility using FACT_RELATIONSHIP. This last part hasn't really been discussed so I welcome feedback here.
Summary