search

cqframework / ecqm-content-qicore-2023

FHIR-based AU 2023 Measures
Creative Commons Zero v1.0 Universal
2 stars 5 forks source link

readme

ecqm-content-qicore-2023

eCQM Measure Content (Using QICore 4.1.1, based on FHIR R4 v4.0.1)

These draft FHIR-based measures and shared libraries are translated from the QDM-based versions of eCQMs to be published in May 2023 for the 2024 reporting year, and have specific versions, especially for the shared libraries, appropriate to the content for that publication update.

Commits to this repository will automatically trigger a build of the continuous integration build, available here:

https://build.fhir.org/ig/cqframework/ecqm-content-qicore-2023

Connectathon Scenarios

Measure Testing

This repository is populated with FHIR-based measure specifications developed using the MADiE tool and based on published AU2023 eCQMs. See the Measure Content Index for a complete listing of available measure content for measure testing.

Multi-Provider Patient Scenario

For systems that deal house data from multiple providers, it is often the case that individual patients may have encounters with multiple providers. In these cases, there must be a way to support identifying the data that should be reported with an individual based on the provider being reported. This is often referred to as the attribution of a patient (and their associated data for reporting) to a provider. Because attribution models vary based on where and why a measure is being evaluated (i.e. for what reporting program), attribution is not included as part of the measure specification. This means that currently, reporting systems must partition data differently when running the same measure for different programs. To support formal description of the attribution model, this scenario will look at ways to describe the attribution as part of the program, rather than as part of the measure itself.

The testing scenario we are using to support these investigations is as follows:

Multi-provider patient scenario

The patient, Jane Doe, has 9 encounters, one at the beginning of each month. The first four encounters are with practitioners Jane Smith and John Smith, practicing at Organization A with CCN CCN-123 and TIN EIN-00123. The last five encounters are with practitioners Kelly Smith and John Smith, practicing at Organization B with CCN CCN-999 and TIN EIN-00999.

To support identifying the provider being reported, use the provider parameter of the R5 $evaluate-measure:

GET [base]/Measure/<measure-id>/$evaluate-measure?subject=Patient/MultiProviderPatient&periodStart=2023-01-01&periodEnd=2023-12-31&provider=Organization/organization-a

This provides the needed input to the attribution criteria to determine what data should be attributed to the provider for the given patient.

There are two test measures defined to support this scenario testing, a "hospital" encounter-based measure MPPEncounterLevel, and a "clinic" patient-based measure MPPPatientLevel.

In addition, there is an AttributionModel library that introduces two fluent functions, isAttributable(Patient) and isAttributable(Encounter):

parameter "Measurement Period" Interval<DateTime> default Interval[@2023-01-01T00:00:00.0Z, @2023-12-31T23:59:59.999Z]
parameter "Provider" String

context Patient

define fluent function isAttributable(patient Patient):
  "Provider" is not null implies Patient.managingOrganization.reference.endsWith("Provider")

define fluent function isAttributable(encounter Encounter):
  encounter.period during "Measurement Period"
    and "Provider" is not null implies encounter.serviceProvider.reference.endsWith("Provider")

These functions are then referenced in the initial population criteria for each of these measures:

define "Initial Population":
  "Encounter with Principal Diagnosis and Age" Encounter
    where Encounter.isAttributable()
define "Has Qualifying Encounter During First 240 Days of Measurement Period":
  exists ( ( ["Encounter": "Office Visit"]
      union ["Encounter": "Outpatient Consultation"]
      union ["Encounter": "Annual Wellness Visit"]
      union ["Encounter": "Face-to-Face Interaction"]
      union ["Encounter": "Home Healthcare Services"]
      union ["Encounter": "Preventive Care Services Established Office Visit, 18 and Up"]
      union ["Encounter": "Preventive Care Services Initial Office Visit, 18 and Up"]
      union ["Encounter": "Preventive Care Services, Initial Office Visit, 0 to 17"]
      union ["Encounter": "Preventive Care, Established Office Visit, 0 to 17"]
      union ["Encounter": "Telephone Visits"]
    ) QualifyingEncounter
      where QualifyingEncounter.period during day of Interval[start of "Measurement Period", start of "Measurement Period" + 240 days]
        and QualifyingEncounter.isAttributable()
  )

define "Initial Population":
  "Has Qualifying Encounter During First 240 Days of Measurement Period"

The "default" attribution model defined by this library is simply that the Patient's managing organization matches the given Provider, and that the Encounter's serviceProvider matches the given Provider.

Whenever measure intent requires the determination of attribution (usually in the qualifying encounter, but sometimes in partitioning encounters to be considered as part of the measure logic below population criteria), these functions can be used to include attribution consideration, but in a way that keeps the attribution model used isolated from the rest of the measure logic.

Subsequently, different attribution logic can be provided by specifying a different version of the AttributionModel library using an artifact manifest:

{
  "resourceType": "Library",
  "id": "MPPManifest",
  "contained": [{
    "resourceType": "Parameters",
    "id": "expansion-parameters",
    "parameter": [{
      "name": "forceCanonicalVersion",
      "valueCanonical": "http://ecqi.healthit.gov/ecqms/Library/AttributionModel|1.0.0-custom"
    }]
  }],
  "extension": [{
    "url": "http://hl7.org/fhir/us/cqfmeasures/StructureDefinition/cqfm-expansionParameters",
    "valueReference": {
      "reference": "#expansion-parameters"
    }
  }],
  "url": "http://ecqi.healthit.gov/ecqms/Library/MPPManifest",
  "name": "MPPManifest",
  ...
}

NOTE: May want to keep expansion parameters limited to only those parameters that would be supplied to the $expand operation, and provide a separate set of parameters for "resolutionParameters"?

NOTE: May want to expand this beyond just version override to actual canonical reference override?

In this way, the measure specification itself can remain unchanged and computable descriptions of different attribution models can be provided.

Terminology API Testing

To continue verifying terminology service capabilities as described by the Measure Terminology Service in the Quality Measure IG. Specifically, validating that:

  1. The service supports read and search for CodeSystem and ValueSet resources
  2. The service supports hosted as well as authored content for ValueSet resources
  3. The service returns resources that conform to the Shareable/Publishable/Computable/Executable profiles for ValueSet resources
  4. The service supports the use of manifest libraries to provide expansion parameters
  5. The service supports the $expand operation with manfiest libraries

The current set of tests is available here:

https://github.com/HL7/cqf-measures/tree/master/thunder-tests

In particular, we are testing the expansion of a grouping value set that requires versions of the component value sets:

Cancer ValueSet 2.16.840.1.113883.3.526.3.1010 (2020-03-06)

This version of the value set references two component value sets:

ValueSet Name
2.16.840.1.113883.3.526.2.1078 Cancer (ICD-10)
2.16.840.1.113883.3.526.2.1079 Cancer (SNOMED)

Each of these value sets has multiple versions as they have changed over time:

ValueSet Version
2.16.840.1.113883.3.526.2.1078 (Cancer ICD-10) 2019-03-15
2021-02-18
2022-02-18
ValueSet Version
2.16.840.1.113883.3.526.2.1079 (Cancer SNOMED) 2019-03-15
2020-03-06
2021-02-18
2022-02-18
2023-02-17

In the eCQM Update 2022-05-05 release, this value set was expanded with the following binding parameters:

ValueSet CodeSystem (Version)
2.16.840.1.113883.3.526.2.1078 ICD10CM (2022)
2.16.840.1.113883.3.526.2.1079 SNOMEDCT (2021-09, 2021-03, 2020-09, 2020-03, 2019-09, 2019-03, 2018-09, 2018-03, 2017-09, 2017-03, 2016-09, 2015-09)

In the eCQM Update 2023-05-04 release, this value set was expanded with the following binding parameters:

ValueSet CodeSystem (Version)
2.16.840.1.113883.3.526.2.1078 ICD10CM (2023)
2.16.840.1.113883.3.526.2.1079 SNOMEDCT (2022-09, 2022-03, 2021-09, 2021-03, 2020-09, 2020-03, 2019-09, 2019-03, 2018-09, 2018-03, 2017-09, 2017-03, 2016-09, 2015-09)

If we just request an expansion of the value set as follows:

https://uat-cts.nlm.nih.gov/fhir/res/ValueSet/$expand?url=http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.3.1010&_format=json

The response is an expansion using the latest version of the value set, the latest version of the component value sets, and the latest version of the code systems used:

2.16.840.1.113883.3.526.3.1010-latest-expansion

However, if we ask for the expansion with expansion identifier eCQM Update 2022-05-05:

https://uat-cts.nlm.nih.gov/fhir/res/ValueSet/$expand?url=http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.3.1010&expansion=eCQM%20Update%202022-05-05&_format=json

The response is an expansion with the expansion profile parameters as established by the eCQM Update 2022-05-05 release:

2.16.840.1.113883.3.526.3.1010-eCQM-Update-2022-05-05-expansion

Similarly, if we ask for the expansion with expansion identifier eCQM Update 2023-05-04:

https://uat-cts.nlm.nih.gov/fhir/res/ValueSet/$expand?url=http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.3.1010&expansion=eCQM%20Update%202023-05-04&_format=json

The response is an expansion with the expansion profile parameters as established by the eCQM Update 2023-05-04 release:

https://uat-cts.nlm.nih.gov/fhir/res/ValueSet/$expand?url=http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.3.1010&expansion=eCQM%20Update%202023-05-04&_format=json

2.16.840.1.113883.3.526.3.1010-eCQM-Update-2023-05-04-expansion

Note that this expansion includes an expansion parameter that is a reference to a manifest:

{
  "name": "manifest",
  "valueString": "http://cts.nlm.nih.gov/fhir/Library/eCQM Update 2023-05-04"
}

Note that the manifest value here is supposed to be a canonical, and encoding it as such and searching for libraries with that URL gives:

https://uat-cts.nlm.nih.gov/fhir/res/Library?url=http://cts.nlm.nih.gov/fhir/Library/eCQM%20Update%202023-05-04

Although this does result in a library, the returned library does not have a url element. The expected result is a library manifest that provides the appropriate code system and value set versions for use in expanding value sets as part of the release:

Library/ecqm-update-2023-05-04

This library contains an expansionParameters extension that references a contained Parameters resource with the relevant expansion parameters:

{
    "resourceType": "Parameters",
    "id": "expansion-parameters",
    "parameter": [{
        "name": "activeOnly",
        "valueBoolean": false
    }, {
        "name": "system-version",
        "valueCanonical": "http://hl7.org/fhir/sid/icd-10-cm|2022"
    }, {
        "name": "system-version",
        "valueCanonical": "http://snomed.info/sct|http://snomed.info/sct/731000124108/version/20220901"
    }, {
        "name": "canonicalVersion",
        "valueCanonical": "http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.3.1010|20200306"
    }, {
        "name": "canonicalVersion",
        "valueCanonical": "http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.2.1078|20220218"
    }, {
        "name": "canonicalVersion",
        "valueCanonical": "http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.2.1079|20230217"
    }]
}

In this case, the expansion parameters indicate:

Note that some terminology services use the concept of an expansion profile to establish these parameters for a particular release. The manifest approach described here is making those parameters explicit as part of the definition of a release, rather than relying on a terminology-server-specific construct to provide this capability.

To make this explicit, the CQF Measures IG defines an extension to the ValueSet/$expand operation to support specifying value set versions:

CQFM ValueSet Expand

Using these parameters in the ValueSet/$expand directly would look like:

[GET] https://uat-cts.nlm.nih.gov/fhir/res/ValueSet/$expand?url=http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.3.1010&valueSetVersion=20200306&activeOnly=false&system-version=http%3A%2F%2Fhl7.org%2Ffhir%2Fsid%2Ficd-10-cm%7C2022&system-version=http%3A%2F%2Fsnomed.info%2Fsct%7Chttp%3A%2F%2Fsnomed.info%2Fsct%2F731000124108%2Fversion%2F20220901&canonicalVersion=http%3A%2F%2Fcts.nlm.nih.gov%2Ffhir%2FValueSet%2F2.16.840.1.113883.3.526.2.1078%7C20220218&canonicalVersion=http%3A%2F%2Fcts.nlm.nih.gov%2Ffhir%2FValueSet%2F2.16.840.1.113883.3.526.2.1079%7C20230217&_format=json

Rather than having to encode these parameters directly in the expand, the manifest parameter allows them to be specified by reference to a manifest library:

[GET] https://uat-cts.nlm.nih.gov/fhir/res/ValueSet/$expand?url=http://cts.nlm.nih.gov/fhir/ValueSet/2.16.840.1.113883.3.526.3.1010&manifest=http://cts.nlm.nih.gov/fhir/Library/eCQM%20Update%202023-05-04&_format=json

This formulation should result in the same expansion as specifying the parameters directly in the $expand.

ClaimsData Testing

To support investigation of the use of Administrative/Claims Data for use in quality reporting, we have a real-world certified to be HIPAA compliant anonymized data set of claims data from a national payer. The data set has about 740 patients, conforming to the CARIN BlueButton IG

There are at least two potential areas of investigation:

  1. Make use of the data here directly from CQL
  2. Create synthetic clinical data derived from the information in the claims data

Artifact Manifest Testing

Measure Narrative Review

Authoring

To author content in this implementation guide, you can use the VS Code CQL Plugin. This plugin will enable you to author, validate, and execute FHIR-based eCQM content.

For instructions on setting up your environment to use the VS Code Plugin, see the VS Code CQL Support Readme

Github Codespaces can also be used in this repository.

Content Indexes

Repository Structure

This repository is setup like any HL7 FHIR IG project but also includes the CQL files and test data which means the file structure will be as follows:

   |-- _genonce.bat
   |-- _genonce.sh
   |-- _refresh.bat
   |-- _refresh.sh
   |-- _updatePublisher.bat
   |-- _updatePublisher.sh
   |-- _updateCQFTooling.bat
   |-- _updateCQFTooling.sh
   |-- ig.ini
   |-- bundles
       |-- MAT
           |-- <bundle files>
       |-- measure
           |-- <bundles>
   |-- input
       |-- ecqm-content-qicore-2023.xml
       |-- cql
           |-- <library name>.cql
       |-- pagecontent
       |-- resources
           |-- library
               |-- <library name>.json
           |-- measure
               |-- <measure name>.json
       |-- tests
           |-- measure
               |-- <measure name>
                   |-- <test case patient id>
                       |-- <test case resource files>
                   |-- ...
       |-- vocabulary
           |-- valueset

Extracting MAT Packages

The CQF Tooling provides support for extracting a MAT exported package into the directories of this repository so that the measure is included in the published implementation guide. To do this, place the MAT export files (unzipped) in a directory in the bundles\mat directory, and then run the following tooling command:

[tooling-jar] -ExtractMatBundle bundles\mat\[bundle-directory]\[bundle-file]

For example:

input-cache\tooling-cli-2.4.0.jar -ExtractMATBundle bundles\mat\CLONE124_v6_03-Artifacts\measure-json-bundle.json

Refresh IG Processing

The CQF Tooling provides "refresh" tooling that performs the following functions:

Whenever changes are made to the CQL, Library, or Measure resources, run the _refresh command to refresh the implementation guide content with the new content, and then run _genonce to run the publication tooling on the implementation guide (the same process that the continuous integration build uses to publish the implementation guide when commits are made to this repository).