1 of 8

FHIR Connectathon 16

Clinical Reasoning Track Report

2 of 8

Participants

  • NCQA
  • MiHIN
  • Transcend Insights
  • Lantana
  • Phillips Healthcare
  • Lantana Consulting
  • Global Strategies
  • Apelon

3 of 8

Scenarios

  • Quality Measurement
  • Opioid Management Decision Support

4 of 8

Quality Measurement

  • Starting with fully specified HEDIS 2018 specs for
    • Colorectal, Cervical and Breast Cancer Screening
  • Terminology challenges
    • Don’t have all the value sets needed for eligibility, so had to comment out those sections
  • Single Value Sets vs Grouping Value Sets
    • For example, Cervical Cytology has a value set for CPT, HCPCS, and LOINC
    • Current approach defines a Grouping value set that is the union of all 3
    • Seeking EHR feedback on whether that’s easier or harder than keeping them separate and unioning the result in the measure
  • Working through applying to a population
    • In a payer scenario, start with an eligible population
    • Potentially need different versions of the measure
      • One to run on the provider, which gathers data
      • One to run on the aggregate set, which stratifies by program

5 of 8

6 of 8

Decision Support Evaluation

  • Phillips Healthcare working on expanding capabilities for evaluation of an ActivityDefinition
    • Additional functionality to help enable testing in isolation

7 of 8

Outcomes

  • MiHIN was able to successfully evaluate BCS measure definition against the test data set
  • Using CQL Runner, we were able to successfully evaluate the BCS measure definition, with the engine accessing the Apelon terminology server to retrieve value set expansions
  • Lantana was able to install a local CQF Ruler and evaluate CDC Opioid Guidance with it

8 of 8

Recommendations

  • There are a lot of moving parts in the stack, need better testing and isolation of the components
  • Overall need better preparation, measure specs weren’t fully available until a week before the connectathon, so we were still troubleshooting measure evaluation
  • Need two different flavors of the measure specifications, at least for the payer calculation scenario, one for the provider that is focused on a single EHR/provider information, and another that is focused on aggregate data and stratifies by program