"COSRI" Clinical Opioid Summary with Rx Integration, is based on the "AHRQ Pain Management Summary SMART on FHIR Application"

The COSRI Pain Management Summary tool was developed by making substantial enhancements to the AHRQ CDS Pain Management Summary tool. That tool was developed as part of the CDS Connect project https://cds.ahrq.gov/, sponsored by the Agency for Healthcare Research and Quality (AHRQ), and developed under contract with AHRQ by MITRE's CAMH FFRDC. The CDS Pain Management Summary tool is a SMART on FHIR dashboard for managing pain, intended to be run within an EHR. The development of COSRI on top of that work entailed enhancements including integration of a FHIR endpoint for PDMP data, new user interface elements, new CQL logic representing WA state rules, MED calculations and visualizations, auditing, and other features. Needless to say, we stand on the shoulders of giants, and are grateful that they shared their wisdom and experience through an open source codebase that served us well as a point of departure for developing COSRI.

"COSRI" is developed by https://cirg.washington.edu as part of our open-source body of work (https://github.com/uwcirg) for the Washington State Department of Health and Washington State Health Care Authority.

About (from AHRQ CDS description, updated)

The Pain Management Summary SMART on FHIR application was developed to support the pilot of the CDS artifact, Factors to Consider in Managing Chronic Pain: A Pain Management Summary. This artifact presents a variety of key "factors" for clinicians to consider when assessing the history of a patient's chronic pain. These factors include subjective and objective findings, along with recorded treatments and interventions to inform shared decision making on treatments moving forward.

The Pain Management Summary SMART on FHIR application was piloted during Summer 2018. Local modifications and development were needed to fully support this application in the pilot environment. For example, custom development was needed to expose pain assessments via the FHIR API. See the pilot reports for more information.

This application was originally piloted with support for FHIR DSTU2. The app has been updated since the pilot to also support FHIR R4, although pilot R4 support has not been piloted in a clinical setting and COSRI production implementations now user R4.

This prototype application is part of the CDS Connect project, sponsored by the Agency for Healthcare Research and Quality (AHRQ), and developed under contract with AHRQ by MITRE's CAMH FFRDC.

Contributions

For information about contributing to this project, please see CONTRIBUTING.

Development Details

The Pain Management Summary is a web-based application implemented with the popular React JavaScript framework. The application adheres to the SMART on FHIR standard, allowing it to be integrated into EHR products that support the SMART on FHIR platform. To ensure the best adherence to the standard, the Pain Management Summary application uses the open source FHIR client library provided by the SMART Health IT group.

The logic used to determine what data to display in the Pain Management Summary is defined using CQL and integrated into the application as the corresponding JSON ELM representation of the CQL. The application analyzes the JSON ELM representation to determine what data is needed and then makes the corresponding queries to the FHIR server.

Once the necessary FHIR data has been retrieved from the EHR, the open source CQL execution engine library is invoked with it and the JSON ELM to calculate the structured summary of the data to display to the user. This structured summary is then used by the React components to render a user-friendly view of the information.

Limitations

This CDS logic queries for several concepts that do not yet have standardized codes. To support this, the following local codes have been defined:

Code	System	Display
PEGASSESSMENT	http://cds.ahrq.gov/cdsconnect/pms	Pain Enjoyment General Activity (PEG) Assessment
PEGPAIN	http://cds.ahrq.gov/cdsconnect/pms	Pain
PEGENJOYMENT	http://cds.ahrq.gov/cdsconnect/pms	Enjoyment of life
PEGGENERALACTIVITY	http://cds.ahrq.gov/cdsconnect/pms	General activity
STARTBACK	http://cds.ahrq.gov/cdsconnect/pms	STarT Back Screening Tool
SQETOHUSE	http://cds.ahrq.gov/cdsconnect/pms	Single question r/t ETOH use
SQDRUGUSE	http://cds.ahrq.gov/cdsconnect/pms	Single question r/t drug use
MME	http://cds.ahrq.gov/cdsconnect/pms	Morphine Milligram Equivalent (MME)

Systems integrating the Pain Management Summary will need to expose the corresponding data as observations using the codes above. As standardized codes become available, these local codes will be replaced.

To build and run in development:

1. Install Node.js (LTS edition, currently 8.x)
2. Install Yarn (1.3.x or above)
3. Install dependencies by executing yarn from the project's root directory
4. If you have a SMART-on-FHIR client ID, edit public/launch-context.json to specify it
5. NOTE: The launch context contains "completeInTarget": true. This is needed if you are running in an environment that initializes the app in a separate window (such as the public SMART sandbox). It can be safely removed in other cases.
6. If you'll be launching the app from an Epic EHR, modify .env to set REACT_APP_EPIC_SUPPORTED_QUERIES to true
7. Serve the code by executing yarn start (runs on port 8000)

To build and deploy using a standard web server (static HTML and JS)

The Pain Management Summary can be deployed as static web resources on any HTTP server. There are several customizations, however, that need to be made based on the site where it is deployed.

1. Install Node.js (LTS edition, currently 8.x)
2. Install Yarn (1.3.x or above)
3. Install dependencies by executing yarn from the project's root directory
4. Modify the homepage value in package.json to reflect the path (after the hostname) at which it will be deployed a. For example, if deploying to https://my-server/pain-mgmt-summary/, the homepage value should be "http://localhost:8000/pain-mgmt-summary" (note that the hostname need not match) b. If deploying to the root of the domain, you can leave homepage as "."
5. Modify the clientId in public/launch-context.json to match the unique client ID you registered with the EHR from which this app will be launched
6. NOTE: The launch context contains "completeInTarget": true. This is needed if you are running in an environment that initializes the app in a separate window (such as the public SMART sandbox). It can be safely removed in other cases.
7. If you've set up an analytics endpoint (see below), set the analytics_endpoint and x_api_key in public/config.json
8. If you'll be launching the app from an Epic EHR, modify .env to set REACT_APP_EPIC_SUPPORTED_QUERIES to true a. This modifies some queries based on Epic-specific requirements
9. Run yarn build to compile the code to static files in the build folder
10. Deploy the output from the build folder to a standard web server

Optionally to step 9, you can run the static build contents in a simple Node http-server via the command: yarn start-static.

To run the unit tests

To execute the unit tests:

1. Run yarn test

To test the app using the public SMART sandbox

Run the app via one of the options above, then:

1. Browse to http://launch.smarthealthit.org/
2. Select R2 (DSTU2) or R4 from the FHIR Version dropdown
3. In the App Launch URL box at the bottom of the page, enter: http://localhost:8000/launch.html
4. Click Launch App!
5. Select a patient

To upload test patients to the public SMART sandbox

Testing this SMART App is more meaningful when we can supply test patients that exercise various aspects of the application. Test patients are represented as FHIR bundles at src/utils/dstu2_test_patients and r4_test_patients. To upload the test patients to the public SMART sandbox:

1. Run yarn upload-test-patients

This adds a number of patients, mostly with the last name "Jackson" (for example, "Fuller Jackson" has entries in every section of the app). The SMART sandbox may be reset at any time, so you may need to run this command again if the database has been reset.

To test the app in standalone mode using the public SMART sandbox

The SMART launcher has a bug that doesn't allow IE 11 to enter the launch URL. This makes testing in IE 11 very difficult. To overcome this, you can reconfigure the app as a standalone app. To do so, follow these steps:

1. Overwrite the /public/launch-context.json file with these contents:

   {
     "clientId": "6c12dff4-24e7-4475-a742-b08972c4ea27",
     "scope":  "patient/*.read launch/patient",
     "iss": "url-goes-here"
   }

2. Restart the application server
3. Browse to http://launch.smarthealthit.org/
4. Select R2 (DSTU2) or R4 from the FHIR Version dropdown
5. In Launch Type, choose Provider Standalone Launch
6. Copy the FHIR URL in the FHIR Server URL box at the bottom of the page (e.g., http://launch.smarthealthit.org/v/r2/sim/eyJoIjoiMSIsImkiOiIxIiwiaiI6IjEifQ/fhir)
7. Paste it into /public/launch-context.json file where url-goes-here is
8. Browse to http://localhost:8000/launch.html

NOTE: Do not check in the modified launch-context.json!

To test the app using the Epic SMART sandbox

The public Epic sandbox does not provide any synthetic patients that exercise the Pain Management Summary logic very well. For this reason, testing against the public Epic sandbox is generally only useful to prove basic connection capability.

Run the app via one of the options above, then:

1. Browse to https://open.epic.com/Launchpad/Oauth2Sso
2. Select a patient from the dropdown
3. In the YOUR APP'S LAUNCH URL box, enter: http://localhost:8000/launch.html
4. In the YOUR APP'S OAUTH2 REDIRECT URL box, enter: http://localhost:8000/
5. Click Launch App

Advanced: To post application analytics

This app can post JSON-formatted analytic data to an endpoint each time the application is invoked.

The data that is posted reports whether or not the patient met the CDS inclusion criteria, lists each section and subsection of the summary (along with the number of entries in each subsection), and provides an overall count of entries. The basic form of the data is as follows:

{
  "meetsInclusionCriteria": <boolean>,
  "sections": [
    {
      "section": <stringName>,
      "subSections": [
        { "subSection": <stringName>, "numEntries": <intCount> },
        ...
      ]
    },
    ...
  ],
  "totalNumEntries": <intCount>
}

To enable posting of analytics, configure the analytics_endpoint and x_api_key in the public/config.json file. The default value is an empty string, which will not post any analytics.