ecqm-bundler

CLI for bundling CQL files as an eCQM FHIR Bundle.

• Basic Measure Bundle Generation from CQL
  • Customizing Population Expressions
  • Dealing with CQL Dependencies
  • Customizing Measure Properties
  • Supplemental Data Elements
  • Risk Adjustment Variables
  • ValueSet Resolution
  • Customizing Canonical URLs
  • Bundling from ELM Content
• Advanced Usage
  • Interactive Mode
  • Multiple Initial Populations
  • Multiple Measure Observations
  • Multiple Measure Groups
  • Composite Measures

Installation

The CLI can be globally installed through npm:

npm install -g ecqm-bundler

Usage

ecqm-bundler can do the following:

• Bundle a new measure from CQL or ELM (most common)
  • ecqm-bundler generate --help
• Combine the groups of existing measure bundles into one
  • ecqm-bundler combine-groups --help
  • See section on combining measure groups
• Create a composite measure from existing measure bundles
  • ecqm-bundler make-composite --help
  • See section on Composite Measures

For bundling with CQL files as input, you must have an instance of the cql-translation-service running somewhere, e.g.:

docker run -d -p 8080:8080 cqframework/cql-translation-service:latest

Bundling is also supported with JSON ELM content directly. See Bundling from ELM Content

Basic Usage

ecqm-bundler generate -c /path/to/main/cql/file.cql -v /path/to/valueset/directory --ipop <ipp-cql-expression> --numer <numer-cql-expression> --denom <denom-cql-expression>

NOTE: Based on the scoring code provided, the CLI will enforce the constraints listed in Table 3-1 of the cqf measures IG. This means that you must specify the minimum valid population expressions for your Measure's scoring type.

This behavior can be disabled using the --disable-constraints CLI option.

Customizing Population Expressions

The bundler will add population group criteria to the Measure resource, which references specific CQL/ELM expressions that identify the relevant eCQM population. These can be customized with the following CLI options:

--ipop <expr>                      "initial-population" expression name of measure
--numer <expr>                     "numerator" expression name of measure
--numex <expr>                     "numerator-exclusion" expression name of measure
--denom <expr>                     "denominator" expression name of measure
--denex <expr>                     "denominator-exclusion" expression name of measure
--denexcep <expr>                  "denominator-exception" expression name of measure
--msrpopl <expr>                   "measure-population"  expression name of measure
--msrpoplex <expr>                 "measure-population-exclusion" expression name of measure
--msrobs <expr>                    "measure-observation" expression name of measure

ecqm-bundler generate -c /path/to/main/cql/file.cql --deps-directory /path/to/deps/directory -v /path/to/valueset/directory --numer "numer def" --denom "denom def" --ipop "ipop def"

Dependencies

If your CQL depends on other cql (i.e. it uses an include <otherlib> ... statement, that CQL must be passed in to the CLI as well via either the --deps or --deps-directory arguments:

Individual Dependency List

ecqm-bundler generate -c /path/to/main/cql/file.cql --deps /path/to/dep1.cql /path/to/dep2.cql -v /path/to/valueset/directory --ipop <ipp-cql-expression> --numer <numer-cql-expression> --denom <denom-cql-expression>

Dependency Directory

ecqm-bundler generate -c /path/to/main/cql/file.cql --deps-directory /path/to/deps/directory -v /path/to/valueset/directory --ipop <ipp-cql-expression> --numer <numer-cql-expression> --denom <denom-cql-expression>

NOTE: It is okay for the dependency directory to include more CQL files that won't be used. The bundler will parse the CQL content to identify all of the libaries that need to be included with the translation request, and will omit any CQL files that aren't included somewhere in the dependency tree starting from the main CQL file

Customizing Measure Properties

Current Measure resource customizations include the Measure Improvement Notation, Population Basis and the Measure Scoring Code:

ecqm-bundler generate -c /path/to/main/cql/file.cql --deps-directory /path/to/deps/directory -v /path/to/valueset/directory --scoring-code proportion --improvement-notation increase --basis boolean <...>

Supplemental Data Elements

To add Supplemental Data Elements (SDEs) to a Measure, simply include expression name(s) using the --sde flag:

ecqm-bundler generate <...> --sde "SDE 1 Expression Name" "SDE 2 Expression Name" "..."

Risk Adjustment Variables

To add Risk Adjustment Variables to a Measure, simply include expression name(s) using the --raf flag:

ecqm-bundler generate <...> --raf "RAF 1 Expression Name" "RAF 2 Expression Name" "..."

ValueSet Resolution

By default, the bundler will look in the directory specified by -v/--valueset for any JSON files that have a url property which matches what is included by any pieces of the CQL logic. You can disable this behavior with the --no-valuesets option, but note that the resulting eCQM FHIR Bundle may not yield proper calculation results due to lack of ValueSet resources:

ecqm-bundler generate -c /path/to/main/cql/file.cql --deps-directory /path/to/deps/directory --no-valuesets <...>

Customizing Canonical URLs

By default, the bundler just uses an example.com URL as the base canonical URL for the resources (e.g. http://example.com/Measure/measure-123). This can be customized using the --canonical-base option:

ecqm-bundler generate -c /path/to/main/cql/file.cql --deps-directory /path/to/deps/directory -v /path/to/valueset/directory --canonical-base "http://example.com/other/canonical/base" <...>

Bundling from ELM Content

ecqm-bundler generate -e /path/to/main/elm/file.json --deps-directory /path/to/deps/directory -v /path/to/valueset/directory <...>

This will forego the CQL translation and bundle the libraries with the ELM content provided.

Advanced Usage/Features

:warning: Highly experimental. Use only if you know exactly what you're doing and why :warning:

Interactive Mode

NOTE: Currently only supported when using CQL as the input

The ecqm-bundler CLI also allows for more complex configuration of a measure bundle using the "interactive" mode:

ecqm-bundler generate <...> --interactive

This will create an interactive prompt where you can construct a more complex measure. For example:

• Measures with multiple population groups
• Measures with multiple initial populations
  • Allows for specifying which initial population the numerator or denominator draw from
• Measures with multiple measure observations
  • Allows for specifying which population an observation references

Interactive mode is highly recommended for constructing complex measures. If you are in an environment where keyboard input is not an option (e.g. in a script), continue to the below sections

Multiple Initial Populations

NOTE: This is only allowed when using --scoring-code ratio

The CLI supports multiple initial populations by passing in multiple values to the --ipop flag

ecqm-bundler generate <...> --scoring-code ratio --ipop ipp1 ipp2 <...>

In the case of multiple initial populations, the numerator and denominator populations must specify which initial population they draw from. This can be done with the --numer-ipop-ref and --denom-ipop-ref options respectively.

IMPORTANT: The values for these flags must match one of the expressions used with the --ipop flag. Otherwise, the CLI will throw an error.

ecqm-bundler generate <...> --scoring-code ratio --ipop ipp1 ipp2 --numer numer --numer-ipop-ref ipp1 --denom denom --denom-ipop-ref ipp2

Multiple Measure Observations

In the case of multiple measure observations, use the --detailed-msrobs flag. The CLI accepts a string of the format <observation-function-name>|<observing-population-expression>. This allows for n many measure observations that reference any population that has already been provided.

IMPORTANT: The values for <observing-population-expression> must match one of the expressions provided with any of the other population expression CLI flags.

ecqm-bundler generate --ipop ipp --numer numer --denom denom --detailed-msrobs "obs1|numer" "obs2|denom"

The above command would generate a measure group where there are two measure observations: obs1, which is a CQL function that observes results from the numer population, and obs2, which is a CQL function that observes results from the denom population

Multiple Measure Groups

:warning: Only use this feature if interactive mode is not a viable solution due to limitations of the environment that ecqm-bundler is being used in :warning:

:warning: Please be very careful if you use this feature :warning:

Due to limitations of command line interfaces, generating a measure with n many groups in one invocation is not feasible. To solve this, ecqm-bundler also comes with a combine-groups command that takes the groups of previously created bundles and combines them into one

:warning: This assumes that the only difference amongst the bundles is the measure group. Everything else (valuesets, library names, etc.) must be exactly the same for this feature to work properly :warning:

ecqm-bundler combine-bundles /path/to/bundle1.json /path/to/bundle2.json /path/to/bundle3.json ... --out combined.json

This will generate a new bundle combined.json where the groups of the measure resources in bundle1.json, bundle2.json, and bundle3.json are combined into one measure group array.

Usage: ecqm-bundler combine-groups [options] <path...>

Combine the groups in the measure resources of two different bundles into one

Options:
  --output <path>  Path to output file (default: "./combined.json")
  -h, --help       display help for command

Composite Measures

:warning: Highly experimental. Use only if you know exactly what you're doing and why :warning:

Similar to combining Measure groups, ecqm-bundler comes with a make-composite command that will generate a Composite Measure that composes existing bundles already created with the generate command

ecqm-bundler make-composite /path/to/bundle1.json /path/to/bundle2.json /path/to/bundle3.json ... --composite-scoring <scoring> --out composite.json

This will generate a new bundle composite.json whose main measure is a composite measure that references the measures that it is composed of in the relatedArtifact list

To customize the group ID or weight of one of the components, use the --detailed-component option, and be sure to follow the proper string format. For example:

ecqm-bundler make-composite <...> --detailed-component "MyMeasure#group-1" # just measure and group
ecqm-bundler make-composite <...> --detailed-component "MyMeasure|1.0.0#group-1" # measure version and group
ecqm-bundler make-composite <...> --detailed-component "MyMeasure#group-1#0.1" # measure group and weight
ecqm-bundler make-composite <...> --detailed-component "MyMeasure##0.1" # just measure and weight

Note that the measure identifying info is required, but group ID and weight are both optional.

:warning: Use of this option will override the autodetection of canonicals that happens in the component measures. Please ensure that the measure identifying info you provide matches the canonical in the corresponding measure resource :warning:

Usage: ecqm-bundler make-composite [options] <path...>

Combine a set of measures into a composite measure

Options:
  --composite-scoring <scoring>  Composite scoring method of the measure (choices: "opportunity", "all-or-nothing", "linear", "weighted", default: "all-or-nothing")
  --detailed-component <component...>  Specify measure components with specific group IDs or weights. Format "measureId(|version)?(#groupId)?(#weight)?" (e.g. "measure|1.0.0#group-2#0.1") (default: [])
  -h, --help                     display help for command

Debugging

Debug mode will write all of the ELM content to a file in the ./debug directory, which it will create. This is useful for inspecting the contents of translated CQL before it gets base64 encoded onto a FHIR Library resource.

To enable, use --debug as an option in the CLI amongst the other options

Full List of Options

For all commands, the following options are shared:

Usage: ecqm-bundler [options] [command]

Options:
  -V, --version                       output the version number
  -o, --out <path>                    Path to output file (default: "./measure-bundle.json")
  --canonical-base <url>              Base URL to use for the canonical URLs of library and measure resources (default: "http://example.com")
  --measure-version <version>         Version of the measure resource
  --improvement-notation <notation>   Measure's improvement notation (choices: "increase", "decrease", default: "increase")
  -h, --help                          display help for command

For the subcommands:

Usage: ecqm-bundler generate [options]

Options:
  -i, --interactive               Create Bundle in interactive mode (allows for complex values) (default: false)
  -c, --cql-file <path>
  -e,--elm-file <path>
  --debug                         Enable debug mode to write contents to a ./debug directory (default: false)
  --deps <deps...>                List of CQL or ELM dependency files of the main file (default: [])
  --deps-directory <path>         Directory containing all dependent CQL or ELM files
  --ipop <expr...>                "initial-Population" expression name(s) of measure (enter multiple values for a multiple ipp ratio measure)
  --numer <expr>                  "numerator" expression name of measure
  --numer-ipop-ref <expr>         expression name of the "initial-population" that the numerator draws from
  --numex <expr>                  "numerator-exclusion" expression name of measure
  --denom <expr>                  "denominator" expression name of measure
  --denom-ipop-ref <expr>         expression name of the "initial-population" that the denominator draws from
  --denex <expr>                  "denominator-exclusion" expression name of measure
  --denexcep <expr>               "denominator-exception" expression name of measure
  --msrpopl <expr>                "measure-population"  expression name of measure
  --msrpoplex <expr>              "measure-population-exclusion" expression name of measure
  --msrobs <expr...>              "measure-observation" expression name of measure (enter multiple values for a measure with multiple observations)
  --detailed-msrobs <expr...>     Specify measure-observation(s) that reference another population. Must be of the format "<observation-function-name>|<observing-population-expression>" (default: [])
  -v, --valuesets <path>          Path to directory containing necessary valueset resource
  --no-valuesets                  Disable valueset detection and bundling
  -u, --translator-url <url>      URL of cql translation service to use (default: "http://localhost:8080/cql/translator")
  -s, --scoring-code <scoring>    Measure's scoring code (choices: "proportion", "ratio", "continuous-variable", "cohort", "composite", default: "proportion")
  -b, --basis <population-basis>  Measure's population basis (default: "boolean")
  --disable-constraints           Bypass the population constraints defined for the given measure scoring code (useful for debugging)
  -h, --help                      display help for command

Usage: index combine-groups [options] <path...>

Combine the groups in the measure resources of two different bundles into one

Options:
  -h, --help  display help for command

Usage: ecqm-bundler make-composite [options] <path...>

Combine a set of measures into a composite measure

Options:
  --composite-scoring <scoring>  Composite scoring method of the measure (choices: "opportunity", "all-or-nothing", "linear", "weighted", default: "all-or-nothing")
  -h, --help                     display help for command

Simple Example

See the example directory for example inputs and outputs to ecqm-bundler. The example-measure-bundle.json file in this directory was generated using the following command:

ecqm-bundler generate -c example/cql/MainLib.cql --deps-directory example/cql -v example/valuesets --ipop "Initial Population" --denom "Denominator" --numer "Numerator" -o example/example-measure-bundle.json