Skip to content

Latest commit

 

History

History
296 lines (213 loc) · 10.2 KB

blazectl.md

File metadata and controls

296 lines (213 loc) · 10.2 KB
Raw

Evaluate a CQL Measure using the Command Line and Blazectl

This section describes how to evaluate a CQL measure using the command line only.

Checkout the Project

This section assumes, that you have checked out the project and open a command line in its directory.

git clone https://github.com/samply/blaze.git
cd blaze

Install Blazectl

Blazectl is a command line utility that can, among other things, evaluate a CQL measure against a Blaze server available via HTTP. You'll find the installation instructions for your platform in its README.

Run Blaze

If you don't already have Blaze running, you can read about how to do it in Deployment. If you have Docker available just run:

docker run -p 8080:8080 -v blaze-data:/app/data samply/blaze:latest

Import some data

If you just started Blaze without any data, you can import some using the blazectl command:

blazectl upload --server http://localhost:8080/fhir .github/test-data/synthea

Evaluate a Simple Measure Counting all Male Patients

blazectl evaluate-measure docs/cql-queries/gender-male.yml --server http://localhost:8080/fhir

Blazectl will output a MeasureReport resource. The important parts are:

{
  "resourceType": "MeasureReport",
  "group": [
    {
      "population": [
        {
          "code": {
            "coding": [
              {
                "system": "http://terminology.hl7.org/CodeSystem/measure-population",
                "code": "initial-population"
              }
            ]
          },
          "count": 56
        }
      ]
    }
  ]
}

You can use jq to extract the count from the MeasureReport:

blazectl evaluate-measure docs/cql-queries/gender-male.yml --server http://localhost:8080/fhir | jq '.group[0].population[0].count'

Now the output should be:

Evaluate measure with canonical URL urn:uuid:73a6023d-5a91-4822-9e36-48ebc2a8ed09 on http://localhost:8080/fhir ...

56

The Measure YAML

Because evaluating a measure requires one to create a Measure and a Library resource on Blaze and execute the $evaluate-measure operation, Blazectl takes a simplified Measure resource in YAML format. It looks like this:

library: docs/cql-queries/gender-male.cql
group:
- type: Patient
  population:
  - expression: InInitialPopulation
  • you reference the CQL library file under the library key
  • you can specify several groups were you can define populations based on a type. Currently only Patient is possible
  • under population you specify the name of a CQL expression which has to be present in the CQL library file. Here the expression is called InInitialPopulation.

The CQL Library File

The CQL library file looks like this:

library "gender-male"
using FHIR version '4.0.0'
include FHIRHelpers version '4.0.0'

context Patient

define InInitialPopulation:
  Patient.gender = 'male'

You can read more about CQL in the Author's Guide and in the CQL Reference. The important bit here is the expression definition:

define InInitialPopulation:
  Patient.gender = 'male'

In it, the expression Patient.gender = 'male' is given the name InInitialPopulation that can be used in the Measure YAML.

Evaluate a Measure Outputting the Distribution of all Patients Birth Years

The goal is to write a Measure that will output the distribution of all patients birth years including a conversation into the CSV format for easy consumption of analysis tools.

First try the command:

blazectl evaluate-measure docs/cql-queries/stratifier-birth-year.yml --server http://localhost:8080/fhir | jq -rf docs/cql-queries/stratifier-birth-year.jq

The (shortened) output should be:

Evaluate measure with canonical URL urn:uuid:d1826406-431d-4993-ace0-6e5e96ee1b48 on http://localhost:8080/fhir ...

"year","count"
"1919",3
"1929",7
"1939",1
"1942",1
"1943",2
"1944",1
"1945",2
"1946",5
"1947",1
"1948",1

Now lets look at the three parts, the Measure YAML, the CQL library file and the jq file:

The Measure YAML

library: docs/cql-queries/stratifier-birth-year.cql
group:
- type: Patient
  population:
  - expression: InInitialPopulation
  stratifier:
  - code: birth-year
    expression: BirthYear

The first part of the Measure YAML looks exactly like the one above. The new part is the stratifier. Stratifiers group members of populations according to the specified expression. Here the population consists of patients and the expression is called BirthYear. So that stratifier will group patients by their birth years.

The CQL Library File

The CQL library file looks like this:

library "stratifier-birth-year"
using FHIR version '4.0.0'
include FHIRHelpers version '4.0.0'

context Patient

define InInitialPopulation:
  true

define BirthYear:
  year from Patient.birthDate

Here the expression InInitialPopulation is defined as true in order to include all patients. The expression called BirthYear that is used as stratifier will take the birth date from the patient and return the year component of it.

The jq File

The jq file doing the CSV conversation looks like this:

["year", "count"],
(.group[0].stratifier[0].stratum[] | [.value.text, .population[0].count])
| @csv

Here ["year", "count"] represents the CSV header and (.group[0].stratifier[0].stratum[] | [.value.text, .population[0].count]) extracts the value and the count out of the strata of the MeasureReport. The syntax of jq is documented in its manual.

Evaluate a Measure Outputting the Distribution of all Condition Codes

The goal is to write a Measure that will output the distribution of all condition codes including a conversation into the CSV format for easy consumption of analysis tools.

First try the command:

blazectl evaluate-measure docs/cql-queries/stratifier-condition-code.yml --server http://localhost:8080/fhir | jq -rf docs/cql-queries/stratifier-condition-code.jq

The (shortened) output should be:

Evaluate measure with canonical URL urn:uuid:4f8176a7-d37a-486f-9f66-5e3849e7013f on http://localhost:8080/fhir ...

"system","code","display","count"
"http://snomed.info/sct","241929008","Acute allergic reaction",3
"http://snomed.info/sct","75498004","Acute bacterial sinusitis (disorder)",8
"http://snomed.info/sct","10509002","Acute bronchitis (disorder)",63
"http://snomed.info/sct","132281000119108","Acute deep venous thrombosis (disorder)",2
"http://snomed.info/sct","706870000","Acute pulmonary embolism (disorder)",9
"http://snomed.info/sct","67782005","Acute respiratory distress syndrome (disorder)",3
"http://snomed.info/sct","65710008","Acute respiratory failure (disorder)",14
"http://snomed.info/sct","195662009","Acute viral pharyngitis (disorder)",69
"http://snomed.info/sct","7200002","Alcoholism",1

Now lets look at the three parts, the Measure YAML, the CQL library file and the jq file:

The Measure YAML

library: docs/cql-queries/stratifier-condition-code.cql
group:
- type: Condition
  population:
  - expression: InInitialPopulation
  stratifier:
  - code: code
    expression: Code

The key difference to the former Measure YAML files is the group type of Condition. That type defines all populations of this group to consist of Condition resources instead of Patient resources. This also means that the population expressions defined in this group have to return Condition resources instead of a boolean value. We will see that in a moment. The stratifier looks the same as in former files, but its expression Code will not operate on Condition resources and not Patients.

The CQL Library File

The CQL library file looks like this:

library "stratifier-condition-code"
using FHIR version '4.0.0'
include FHIRHelpers version '4.0.0'

context Patient

define InInitialPopulation:
  [Condition]

define function Code(condition FHIR.Condition):
  condition.code

Here we have our InInitialPopulation expression defined as [Condition]. The expression [Condition] retrieves all Condition resources of the current selected patient. Doing so will return the required type of Condition resources from the population expression, we discussed above. If you like to constrain the patients from which Condition resources are returned, you have to do it in the InInitialPopulation expression.

Next we define a function called Code what will extract the code from a Condition resource. Here the expression condition.code is a FHIRPath expression as CQL is a superset of FHIRPath.

The jq File

The jq file doing the CSV conversation looks like this:

["system", "code", "display", "count"],
(.group[0].stratifier[0].stratum[]
  | [.value.coding[0].system, .value.coding[0].code, .value.coding[0].display, .population[0].count])
| @csv

Because we return a CodeableConcept from our stratifier expression Code, the stratum values will contain that CodeableConcepts directly. In the jq script, we extract the system, code and display parts of the first Coding using the expression .value.coding[0].system and others. If the Condition resources use more than one Coding, its easy to add columns with the later Codings. In the Synthea dataset that's not the case.

Evaluate a Measure Outputting the Distribution of all Body Weights

blazectl evaluate-measure docs/cql-queries/stratifier-body-weight.yml --server http://localhost:8080/fhir | jq -rf docs/cql-queries/stratifier-body-weight.jq

The (shortened) output should be:

Evaluate measure with canonical URL urn:uuid:dbb406a8-e72e-47a2-8ce7-c970bb964a9a on http://localhost:8080/fhir ...

"body-weight-value","body-weight-unit","count"
10.1,"kg",3
10.2,"kg",3
10.4,"kg",4
10.7,"kg",1
10.9,"kg",2
100,"kg",11
100.2,"kg",2
100.5,"kg",1
100.7,"kg",9