Creating FHIR Data

This is a quickstart sample application that introduces some basic FHIR and Medplum concepts.

Prerequisites

You will need the following to get started and instructions on how to set this up were covered in the previous article.

• Make sure you have an account and log into Medplum
• Log in and create Client Application
• Save the Client ID and Client Secret, they will be needed to get your sample to run.

These three requirements will need to be in place to connect

const MY_CLIENT_ID = 'MY_CLIENT_ID';
const MY_CLIENT_SECRET = 'MY_CLIENT_SECRET';

Cloning an existing project

If you have an existing project, and want to copy your data to a new environment (for example to a staging environment), this can easily be done using the $clone operation. For more details ee the Projects guide.

Example App - Simple Lab Results Workflow

This example represents a common healthcare workflow, creating an order for lab tests (a.k.a ServiceRequest in FHIR) for patients and when the lab test is complete creating results (a.k.a. Observation and DiagnosticReport in FHIR) that correspond to the original ServiceRequest.

This example will illustrate how to create FHIR object, how to update them, how to link them, and them how to read them back in bulk.

Here is a breakdown of workflow at a high level

• Authenticate with the server using OAuth client credentials flow
• Use FHIR batch request to create a Patient and a ServiceRequest
  • The example will use a conditional to only create the Patient if it does not already exist
  • The example will link the ServiceRequest to the Patient
• Create an Observation and DiagnosticReport resources
• Read back the DiagnosticReport and Observations
  • Use a batch request to read all Observations in one go, versus making multiple requests

Authenticating using OAuth client credentials flow

The client credentials flow is a type of connection that is used to obtain an access token outside the context of the user.

import { MedplumClient, UCUM } from '@medplum/core';
import fetch from 'node-fetch';

const medplum = new MedplumClient({
  baseUrl: 'https://api.medplum.com/',
  fetch: fetch,
});
await medplum.startClientLogin(MY_CLIENT_ID, MY_CLIENT_SECRET);

Creating the Patient and Service Request

Patient and a ServiceRequest sounds simple, but there are several nuances. If the Patient already exists, a new one should not be created. We also need to ensure that the ServiceRequest is linked to the correct patient.

Creating Patient

Creating a Patient if one does not exist uses the conditional create logic in FHIR. In this example, a patient has an Medical Record Number or MRN. If that MRN exists, then a new patient should not be created. In a lab workflow, it is common for a lab to serve patients repeatedly. In this case where there is already a patient in the system, it would be incorrect (and confusing) to make a new patient record.

import { Patient } from '@medplum/fhirtypes';
import { randomUUID } from 'crypto';

// Generate an example MRN (Medical Record Number)
// We will use this in the "conditional create" and "upsert"
const exampleMrn = randomUUID();
const patientData: Patient = {
  resourceType: 'Patient',
  name: [{ given: ['Batch'], family: 'Test' }],
  birthDate: '2020-01-01',
  gender: 'female',
  identifier: [
    {
      system: 'https://namespace.example.health/',
      value: exampleMrn,
    },
  ],
};

// When creating an order, and if you don't know if the patient exists,
// you can use this MRN to check. Use the 'identifier=' search criterion for a
// conditional create: if a resource with the given `identifier` already exists,
// that resource will be returned instead.
const patient = await medplum.createResourceIfNoneExist(patientData, 'identifier=' + exampleMrn);
console.log('Patient record created', patient);

The behavior of the the Patient.identifier field is important to note. Patient.identifier usually has a reference string or URL that describes which system that identifier came from. Identifiers are a concept in FHIR which describe the context in which that identifier is generated, for example, and identifier could be a Social Security Number (SSN) or be created by a health system for their own internal purposes. Here is an example of an identifier scheme for the Australian Healthcare system.

We recommend that providers put documentation of their identifier system online for interoperability purposes.

Performing an Upsert

In some cases, you may want to update the resource in place if it already exists in the system, in addition to creating it if it doesn't exist yet. This is accomplished via a combined "update" and "insert" operation: an "upsert".

Provide the current version of the resource and a search query, similar to createResourceIfNoneExists(). If the search query resolves to a single resource, that resource will be updated. If it does not find a matching resource, one will be created from the given data. If multiple matches are found, an error will be returned: in this case, more specific search criteria are required to unambiguously identify the resource to be updated or created.

Example: Upsert a Patient record

// An "upsert" (i.e. update/insert) will either update the resource in place if it
// already exists, otherwise is will be created.  This is performed in a single,
// transactional request to guarantee data consistency.
const updatedPatient = medplum.upsertResource(patient, 'identifier=' + exampleMrn);
console.log('Patient record updated', updatedPatient);

Create the ServiceRequest

Creating a new ServiceRequest also has some nuance to it. ServiceRequests in this context can be thought of as a "requisition for a lab test" and the ServiceRequest.code specifies what test panel is being ordered. Most labs will have a concept of a test menu and that should indicate which labs should be run for this service request.

Note that there are many fields on the requisition, and filling them in with the right data is crucial. This example is minimal for clarity.

import { createReference } from '@medplum/core';
import { ServiceRequest } from '@medplum/fhirtypes';

const serviceRequestData: ServiceRequest = {
  resourceType: 'ServiceRequest',
  status: 'active',
  intent: 'order',
  subject: createReference(patient), // link this ServiceRequest to the Patient
  code: {
    coding: [
      {
        system: 'https://samplelab.com/tests',
        code: 'SAMPLE_SKU',
      },
    ],
  },
};

const serviceRequest = await medplum.createResource(serviceRequestData);
console.log('Service Request', serviceRequest.id);

If you are using the hosted Medplum service you can see your ServiceRequest objects here. Similarly, you can see Patients here.

Creating the Diagnostic Report

Once the lab test has been completed and the specimens analyzed, it is time to create a diagnostic report - but it is really important to link that diagnostic report back to the Patient and the corresponding ServiceRequest.

To get this to be linked up, you'll need to have the identifiers for the Patient and ServiceRequest that were created in the previous section.

You can then create a diagnostic report using the function below.

Create the Observations

import { Observation } from '@medplum/fhirtypes';

// Create two observations from the array
const observationData: Observation[] = [
  {
    resourceType: 'Observation',
    basedOn: [createReference(serviceRequest)], // Connect this Observation to the ServiceRequest
    subject: createReference(patient), // Connect this Observation to the Patient
    status: 'preliminary',
    code: {
      coding: [
        {
          system: 'https://samplelabtests.com/tests',
          code: 'A1c',
          display: 'A1c',
        },
      ],
    },
    valueQuantity: {
      value: 5.7,
      unit: 'mg/dL',
      system: UCUM,
      code: 'mg/dL',
    },
  },
  {
    resourceType: 'Observation',
    basedOn: [createReference(serviceRequest)], // Connect this Observation to the ServiceRequest
    subject: createReference(patient), // Connect this Observation to the Patient
    status: 'preliminary',
    code: {
      coding: [
        {
          system: 'https://samplelabtests.com/tests',
          code: 'blood_glucose',
          display: 'Blood Glucose',
        },
      ],
    },
    valueQuantity: {
      value: 100,
      unit: 'mg/dL',
      system: UCUM,
      code: 'mg/dL',
    },
  },
];

// Map through the observation data to create all the observations
const observations = await Promise.all(observationData.map(async (data) => medplum.createResource(data)));

for (const observation of observations) {
  console.log('Created Observation', observation.id);
}

Create Report

import { DiagnosticReport } from '@medplum/fhirtypes';

const reportData: DiagnosticReport = {
  resourceType: 'DiagnosticReport',
  basedOn: [createReference(serviceRequest)], // Connect this DiagnosticReport to the ServiceRequest
  subject: createReference(patient), // Connect this DiagnosticReport to the Patient,
  status: 'preliminary',
  code: {
    coding: [
      {
        system: 'https://samplelab.com/testpanels',
        code: 'SAMPLE_SKU',
      },
    ],
  },
  result: observations.map(createReference), // Create an array of references to the relevant observations
};
const report = await medplum.createResource(reportData);
console.log('Created Report', report.id);

This will create a DiagnosticReport that is linked to the ServiceRequest and to the Patient. If you are using hosted Medplum, you can view all DiagnosticReports here.

Conclusion

Hopefully this simple lab workflow, "ordering a lab" and "getting a lab report" was a good beginner illustration on getting started with FHIR. We welcome your feedback. Please feel free to file issues or submit pull requests.

This sample is based on a service where data is hosted on Medplum, but for those who need the data stored on premise, we do support self-hosting the backend.