Uploaded image for project: 'FHIR Specification Feedback'
  1. FHIR Specification Feedback
  2. FHIR-27878

Permit stratifier expression to have any type (Conformance requirement 15)

    XMLWordPrintableJSON

    Details

    • Type: Change Request
    • Status: Applied (View Workflow)
    • Priority: Medium
    • Resolution: Persuasive
    • Specification:
      US Quality Measures (FHIR)
    • Raised in Version:
      0.1
    • Work Group:
      Clinical Quality Information
    • Related Page(s):
      (NA)
    • Resolution Description:
      Hide

      In HQMF, stratifier criteria could only be expressed in the same way as other population criteria and so the type constraint applied, but in FHIR measures, stratifier criteria can be expressed as codes or paths to elements, so the type constraint should be relaxed. (i.e. don't refer to stratifier expression with this conformance requirement).

      Show
      In HQMF, stratifier criteria could only be expressed in the same way as other population criteria and so the type constraint applied, but in FHIR measures, stratifier criteria can be expressed as codes or paths to elements, so the type constraint should be relaxed. (i.e. don't refer to stratifier expression with this conformance requirement).
    • Resolution Vote:
      Ben Hamlin/Howard Strasberg: 22-0-0
    • Change Category:
      Correction
    • Change Impact:
      Compatible, substantive

      Description

      Conformance Requirement 15 on the quality measure IG states:

      1. For patient-based measures, the CQL stratification expression SHALL return a Boolean.
      2. For event-based measures (e.g. episode-of-care), the CQLstratification expression SHALL return a list of events of the same type as the population criteria.

      This change request is for relaxing the requirement 1 to allow stratification expressions to return any type that can be represented in the MeasureReport.group.stratifier.stratum.value] CodeableConcept field.

      See this zulip thread] for an example of how a stratifier can return a coded value (Patient.gender) instead of a boolean, for a Patient-based measure.

      Comment from Bryn on zulip:

      that conformance requirement exists because that's the way that HQMF represented stratifiers. However, the underlying FHIR resources don't have that requirement and allow more flexible specification of stratifier criteria. You raise a very good point that because the underlying resources allow it, should the Quality Measure IG relax that conformance requirement. Do you mind submitting a tracker to that effect to the Quality Measure IG? We are planning on stratifiers and supplemental data being a focus of the upcoming connectathon, and would love to get feedback on that aspect.

       

      For event-based measures, I think requirement 2 can be relaxed as well. Instead of requiring the list of events to have the same type as the population criteria, the list could have any type, as long as the length of the list is equal to the length of the population list, and in the same order.

      To illustrate, here is an example for an event-based cohort measure that stratifies on Encounter.status:

      // cql

      ```

      define "Initial Population": [Encounter] E where E.period during "Measurement Period"

      define "Status Stratifier": "Initial Population" E return E.status

      ```

      // Measure.group

       
      "stratifier": [
      {
      "code":

      { "text": "stratifier-status" }

      ,
      "criteria":

      { "language": "text/cql", "expression": "Status Stratifier" }

      }
      ]

      // MeasureReport.group
      "stratifier": [
      {
      "code": [

      { "text": "stratifier-status" }

      ],
      "stratum": [
      {
      "value":

      { "text": "planned" }

      ,
      "population": [
      {
      "code": {
      "coding": [

      { "code": "initial-population" }

      ]
      },
      "count": 50
      }
      ]
      },
      {
      "value":

      { "text": "triaged" }

      ,
      "population": [
      {
      "code": {
      "coding": [

      { "code": "initial-population" }

      ]
      },
      "count": 50
      }
      ]
      }
      ]
      }
      ]

        Attachments

          Activity

            People

            Assignee:
            bryn.rhodes Bryn Rhodes
            Reporter:
            miriama Miriam Ashton
            Watchers:
            4 Start watching this issue

              Dates

              Created:
              Updated:
              Resolved:
              Vote Date: