Mapping Microbiology Susceptibility into OMOP CDM4 Observations

Implementers
40

Hi,
We are also in the process of mapping laboratory microbiology data onto OMOP CDC v6.0 and have encountered similar issues as the ones discussed here and is other threads. Our use case may be summarized saying that we aim at detailed observing tests performed in a lab network.
As shown below this relates to a several discussion threads for which we have not seen a firm answer yet.
In the situation of micro-organism identification and subsequent antibiotic susceptibilities, our model raises questions similar to what is described by @parisni, @TBanokina and @nzvyagina
It includes a specimen table (eg. Veinous Blood sample) that is linked by fact_relationship to one or more measurement table.

  • It includes a specimen table (eg. Veinous Blood sample) that is linked by fact_relationship to one or more measurement table.
  • A first measurement table stores information about the microorganisms identification tests and system level results. Such as, in the case of a mass spectroscopy malti-tof test,
    measurement_concept_id = OMOP code for LOINC describing the test eg. 76346-6 Microorganism identified in Isolate by MS.MALDI-TOF
    Value_as_concept_id = OMOP code for the concept representing the organism identified
  • This table is linked by fact_relationship to several measurement tables for antibiotic susceptibility tests & results.
    Measurement table storing antibiotic susceptibility results.
    measurement_concept_id = OMOP code for the LOINC describing the antibiotic test, eg. 28-1 Ampicillin [Susceptibility] by Minimum inhibitory concentration (MIC)
    Value_as_number would be the MIC result
    Value_as_concept_id = OMOP code for the concept representing the corresponding category (R/I/S)

This leads to several questions appearing in several discussion threads for which we seek answers

  1. How can we confirm that using both value_as_concept_id and value_as_number to store antibiotic susceptibility results (MIC & category) for the same measurement is allowed ?
    This is also
  1. We have the same issue than @TBanokina concerning Meas.value domain

It is especially challenging to store organisms identification in value_as_concept_id (eg. Escherichia coli) because SNOMED derived codes for organisms are considered as Observation domain (ID = 4011683). To follow the generally accepted paradigm of “LOINC as the question and SNOMED as the answer”, one would need to allow code from SNOMED vocabulary to be also used as value_as_concept_id
So, how can we proceed to modify and clear this vocabulary issue, thus allowing SNOMED vocabulary to be used for value_as_concept_id in the MEASURMENT table?
Note that a related discussion appears in the thread from @Vojtech_Huser (Duplicate standard concepts for value_as_concept_id in OMOP Vocab) showing that data duplication appear between LOINC answers and SNOMED codes.

  1. We agree with @Christian_Reich, connecting specimen and measurement by the fact relationship is not the cleanest of all options. As @parisni and @Alexdavv suggested (see also Link between SPECIMEN and MEASUREMENT) adding specimen_id as a foreign key to the Measurement table would reduce the need for fact_relationship between those two table (Modifier_of_event_id / Modifier_of_field_concept_id)
    How can we help the community to move forward in this direction and make a decision ?

  2. Lastly what is the status of the proposal for a microbiology or culture area table ? (Proposal : new microbiology table · Issue #281 · OHDSI/CommonDataModel · GitHub and Adding Cultures into OMOP v5)
    Is this still a living proposal ? how can we participate to its improvement (if needed) and blessing ?

Considering that options 3 and 4 are likely mutually exclusive.
E. Theron // X. Gansel