OMOP Quality Measurement 9 February 2022

Minutes from last call: OMOP Quality Measurement - 26 January 2022

Recording of this call (access requires passcode):

• https://us02web.zoom.us/rec/share/qxwHMIPwZR07RXCTbgQlFedyOBQaqdhjyk4ArGixhTe7IUcb5aO5Ovqi36XPPVmQ.JYeVECRvZP7tEImL
• Passcode: W.sJ2X#!

Background:

Options for examples to address in the May FHIR Connectathon.

• Oncology concepts - outcomes for prostate cancer - potential process or outcome measures (Sebastian van Sandijk) - may be able to work with existing eCQMs - potentially stroke as well
• Mortality measures - ask Maria about public health mortality measure work - also interest in aligning COVID vaccination status and mortality
• Maryland perspective - suites including mortality, readmission, and several others - these may not be in eCQM/dQM format. In the ambulatory space, there are eCQMs which could be evaluated via Atlas.
• The mapping will be the most significant issue. Will need to work with the OMOP-FHIR mapping group.
• Daveera Gabriel - working on a COVID dataset to OMOP - if that is useful it can be used - looking at co-morbidities using VSAC value sets
  • https://ncats.nih.gov/n3c 
• Paul Nagy indicates his group has looked at Epic ETL to OMP for the vaccinations and conditions
• VSAC content should allow a one-time mapping exercise to the OMOP content to feed the eCQMs.
  • for any eCQM only using VSAC content (from standard code sets) - should be easier to implement - what else within eCQMs isn't using those standards?

Listserve: http://www.hl7.org/myhl7/managelistservs.cfm?ref=nav See FHIR-OMOP on right

Zulip thread: https://chat.fhir.org/#narrow/stream/286658-omop-.2B.20fhir/topic/FHIR.20.2B.20OMOP.20dQM.20.28Measurement.29

Next All-Hands FHIR-OMOP meeting: TBD (To be moved from the previously scheduled Monday, February 14, 2022)

Discussion:

Thomas White introduced James Brash and Qi Yang who have been working with MedStar

To implement eCQMs - represent as Atlas cohorts (numerator or denominator) - James and Qi may be able to help

• Underlying code set mapping VSAC to OHDSI libraries - and roll it up (OHDSI generally is higher level while VSAC is fully enumerated lists) - how to represent eCQM concepts as parents with children and exclusions rather than enumerated lists
  • Caveat: HAPI Terminology servers do not support intensional expression of VS (but does for LOINC)
• Evaluate eCQMs and determine how to represent the logic in an Atlas cohort - opportunity to prototype high-value eCQMs as Atlas cohorts

Daveera Gabriel - reflect CDC co-morbidity conditions for COVID - largest database of de-identified data (COVID use case using VSAC content) and real-world observational data available. There is issue regarding access to data but some might be able to support the use case but it is not on an Atlas stack. Could be represented on an Atlas instance. Prototype Proposed draft paper about use of VSAC data sets with OMOP. Part of N3C - already has conversion to OMOP (PCORNet, I2b2 ACT (Accrual to Clinical Trials), PedsNet, TriNetX and now doing CMS claims data) - harmonizes 5 common data models to OMOP. (previous work provided by Dan Vreeman: comparing some value sets for quality measures, public health reports, and mapping common tests: https://pubmed.ncbi.nlm.nih.gov/24551424/)

VSAC enables intensional (“dynamic”) value set definition for some terminologies (see “Valid Operators for Each Code System" https://www.nlm.nih.gov/vsac/support/authorguidelines/createvs.html#select)

For eCQM need to understand - Currently identify individuals testing positive for COVID - primarily outpatient, 2:1 matching control (I.e., same source not COVID positive).

Criteria for cohort: https://github.com/National-COVID-Cohort-Collaborative/Phenotype_Data_Acquisition/wiki/Latest-Phenotype

Still need the folks who will be able to "convert" the Atlas cohort to FHIR-CQL - potentially with json, perhaps python to fill out the use case.

There is also a CMS FHIR Connectathon in July and the NCQA/HL7 Digital Quality Summit to follow up efforts from the May Connectathon. (July 13-15, 2022)

To bring the cohort back - what are the tasks that have to happen and who does them? This option would close the loop.  (Ideal patient for denominator cohort, then look for the numerator criteria - focus more on observations that require constant refinement - re-evaluate in OHDSI-OMOP world).

• What has to happen to make this scenario occur: Take eCQM definition, build Atlas cohort - execute in FHIR-CQL - assure it matches with the phenotype in the OHDSI repository.

From last week ---- Workflow process to participate in the May FHIR Connectathon

• Topic - run measure cohort on translated phenotype on OHDSI data - pick a clinical topic
• Need mappings/transforms from OHDSI to CQL representation - consider run-time translation? (White-rabbit utility in Github - useful for mapping local data to OMOP)
• Talk with OMOP mapping cohort to determine which has been/will be translated 
• Get FHIR CQL translation running
• Host server
• Participants to build the data

Tools - Atlas to create cohort, Achilles, other evaluate tools https://www.ohdsi.org/software-tools/ 

Reference using synthetic patient data (2008): https://www.ohdsi.org/web/wiki/lib/exe/fetch.php?media=resources:ohdsionfhir_gatech.pdf

We have ways to generate data based on measure definitions.

• Can we take and existing cohort definition from the OMOP phenotype and use it to build test data in FHIR. Use an existing OMOP Phenotype definition and obtain a measure numerator or denominator and build up FHRI data and run. Or, find a source of de-indentified real-world data and run it on those data.
• Another approach is to take a measure and automatically create an OMOP / OHDSI cohort. An existing measure run against OMOP. Automation for the data conversion will require participants who can consider how to automate.

Next steps:

• Contact OMOP-CQL conversion folks to discuss options and willingness to participate in a Connectathon
• Await discussion next meeting for a specific eCQM cohort (Ben Hamlin indicated that Sebastian has a suggestion)
• Daveera will work with her Hopkins group regarding coordination of an AMIA presentation proposal - possibly a theater presentation/demo.

Adjournment - The meeting was adjourned at 11:01 AM ET. The next scheduled meeting will be February 23, 2022.

Attendees