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 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.
For information about contributing to this project, please see CONTRIBUTING.
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.
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.
- Install Node.js (LTS edition, currently 8.x)
- Install Yarn (1.3.x or above)
- Install dependencies by executing
yarn
from the project's root directory - If you have a SMART-on-FHIR client ID, edit
public/launch-context.json
to specify it - If you'll be launching the app from an Epic EHR, modify
.env
to setREACT_APP_EPIC_SUPPORTED_QUERIES
totrue
- Serve the code by executing
yarn start
(runs on port 8000)
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.
- Install Node.js (LTS edition, currently 8.x)
- Install Yarn (1.3.x or above)
- Install dependencies by executing
yarn
from the project's root directory - Modify the
homepage
value inpackage.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/, thehomepage
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 leavehomepage
as"."
- Modify the
client_id
inpublic/launch-context.json
to match the unique client ID you registered with the EHR from which this app will be launched - If you've set up an analytics endpoint (see below), set the
analytics_endpoint
andx_api_key
inpublic/config.json
- If you'll be launching the app from an Epic EHR, modify
.env
to setREACT_APP_EPIC_SUPPORTED_QUERIES
totrue
a. This modifies some queries based on Epic-specific requirements - Run
yarn build
to compile the code to static files in thebuild
folder - 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 execute the unit tests:
- Run
yarn test
Run the app via one of the options above, then:
- Browse to http://launch.smarthealthit.org/
- In the App Launch URL box at the bottom of the page, enter:
http://localhost:8000/launch.html
- Click Launch App!
- Select a patient
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/test_patients
. To upload the test patients to the public SMART sandbox:
- 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.
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:
- Overwrite the
/public/launch-context.json
file with these contents:{ "client": { "client_id": "6c12dff4-24e7-4475-a742-b08972c4ea27", "scope": "patient/*.read launch/patient" }, "server": "url-goes-here" }
- Restart the application server
- Browse to http://launch.smarthealthit.org/
- In Launch Type, choose Provider Standalone Launch
- Copy the FHIR URL in the FHIR Server URL box (e.g.,
http://launch.smarthealthit.org/v/r2/sim/eyJoIjoiMSIsImkiOiIxIiwiaiI6IjEifQ/fhir
) - Paste it into
/public/launch-context.json
file whereurl-goes-here
is - Browse to http://localhost:8000/launch.html
NOTE: Do not check in the modified launch-context.json!
First install the SMART Platform via: https://github.com/smart-on-fhir/installer
Verify it works via the sample apps included with it:
- Browse to http://localhost:9080/
- Sign in using demo/demo
- Pick "SMART DSTU2 Sandbox"
- Click "Growth Chart" app
- Click "Launch"
- Click "Clark, Susan A."
Then run the app via one of the options above and register it via the SMART Platform:
- Browse to http://localhost:9080/ (if not already signed in)
- Sign in using demo/demo (if not already signed in)
- Pick "SMART DSTU2 Sandbox" (if not already signed in)
- Choose "Register Manually"
- Enter these values: a. App Type: Public Client b. App Name: CQL Demo c. App Launch URI: http://localhost:8000/launch.html d. App Redirect URIs: http://localhost:8000/ e. Allow Offline Access: NO f. Patient Scoped App: YES
- Save
- Click the "CQL Demo" app
- Click "Launch"
- Choose a patient
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:
- Browse to https://open.epic.com/Launchpad/Oauth2Sso
- Select a patient from the dropdown
- In the YOUR APP'S LAUNCH URL box, enter:
http://localhost:8000/launch.html
- In the YOUR APP'S OAUTH2 REDIRECT URL box, enter:
http://localhost:8000/
- Click Launch App
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.