HL7.FHIR.UV.VULCAN-SCHEDULE\Use case for integration using CDISC ODM

This page is part of the Clinical Study Schedule of Activities (v1.0.0: STU1) based on FHIR R4. This is the current published version in its permanent home (it will always be available at this URL). For a full list of available versions, see the Directory of published versions

Use case for integration using CDISC ODM

Alignment between the CDISC Operational Data Model (ODM) and the FHIR SoA Model

The CDISC Operational Data Model (ODM) structure is a common model for representing defined, planned and performed activities in Clinical Data Management Systems (CDMS, EDC, etc.) and Clinical Trial Management Systems (CTMS). It is the de facto standard for exchanging data and metadata between clinical data management systems.

Having a standard alignment between the ODM and FHIR will enable rapid and reproducible system builds using standardised interfaces. Whether the initial study design is developed in the ODM/XML formats or using FHIR Resources, each EHR system should then be able to use the study design as part of study start up activities by a Study Builder. There will need to be some augmentation of the process to make best use of transportable concepts (e.g., test codes, procedure codes, etc.) but it is hoped the work here will be able to be used as a core implementation model.

CDISC Operational Data Model (ODM)

The CDISC ODM is a model designed for the transportation and storage of Study Data for Clinical Studies. The landing page for the standard can be found here: https://www.cdisc.org/standards/data-exchange/odm

Note: the CDISC ODM is owned by CDISC. A free CDISC website account is required to access the material - create an account here.

This document is a light touch overview of the ODM - more information can be found on the CDISC GitHub site:

The CDISC repository for the ODM found here.
ODM Schema

What is the ODM

The CDISC Operational Data Model (ODM) is a XML-based model defined for the transport and archival of Clinical Trials Data; it is broadly made up of the following top-level elements

Clinical Data - actual datapoints gathered on study participants
Metadata - definitions used to capture and report the data collected
Administrative Data - information regarding the collection of data (sites, personnel, etc)
Reference Data - common data needed to interpret the collected data

Structures within the ODM have little semantic meaning; it is a format focused on the representation of planned and performed activities so captures relationships between elements moreso than the actual actual definition of the element itself; for example - there is no characterisation of a participant (in contrast to the Patient resource); there is a SubjectData element used to collect captured datapoints for an individual participant. The SubjectData element includes attributes such as an OID as an identifier, but there is no information model that says the OID is an external identifier (eg Subject ID) or some other internal identifier (eg Primary Key from a Subjects table, as an example).

Data collection is driven by Forms; forms group activities by type (ie all the Vital Signs observations for a single visit are usually grouped in a single form, irrespective of multiple planned timepoints for the elements). Individual fields on a form are Items in ODM parlance and have associated metadata pertaining to the allowable content (eg datatype, length, constrained values or allowable units for a value)

Structural Overview

The ODM specification is distributed as an XML document, and as such can be extended through the use of vendor namespaces; examples where this has been done include:

Study Design Model - extensions to cover topics such as workflow, study design concepts
Dataset-XML - extensions to use for transporting datasets

The implementation of the Study Metadata follows a DEF-REF pattern. We define a type of element (eg Visit, Form, Field) and then reference it one or more times within the corresponding element; as an example:

<StudyEventDef OID="SE.SCREENING_VISIT" Name="Screening Visit (Visit 1)" Repeating="No" Type="Scheduled">
  <Description>
    <TranslatedText xml:lang="en">Screening Visit at day -14</TranslatedText>
  </Description>
  <FormRef FormOID="F.DOV" Mandatory="Yes" OrderNumber="1"/>
  <FormRef FormOID="F.DS_IC" Mandatory="Yes" OrderNumber="2"/>
  <FormRef FormOID="F.IE" Mandatory="Yes" OrderNumber="3"/>
  ...
</StudyEventDef>
...
<FormDef OID="F.DOV" Name="Date of Visit" Repeating="No">
	<Description>
			<TranslatedText xml:lang="en-US">Subject Visits consolidates information about the timing of subject visits that is otherwise spread over domains that include the visit variables (VISITNUM and possibly VISIT and/or VISITDY).</TranslatedText>
	</Description>
	<ItemGroupRef ItemGroupOID="IG.DOV" Mandatory="Yes" OrderNumber="1"/>
</FormDef>

The OID is the key here; it serves as the unique id (for a given –Def element type); the Reference to this –Def in the –Ref element is usually qualified with the Element type; eg

OID are the best key to share data with clinical research systems.

How the schedule of activities implemented in ODM

The linking of Data Elements (outcomes of Observations) and Definitions is illustrated here:

The top-level element for the definitions for the Study are contained within the MetaDataVersion element

MetaDataVersion

The MetaDataVersion is a set of study definition configuration; studies can have multiple versions of a study definition (sometimes in parallel); the version is identified by the OID

<MetaDataVersion Description="LZZT study design version 1" Name="LZZT study design version 1" OID="LZZT_1">
  <Protocol>
    <Description>
      <TranslatedText xml:lang="en">A randomized, double-blind, parallel (3 arm), placebo-controlled trial of 26 weeks duration.</TranslatedText>
    </Description>
...
</MetaDataVersion>

Protocol

The Protocol lists the study events/encounters that can occur within a Study.

<Protocol>
	<Description>
		<TranslatedText xml:lang="en">A randomized, double-blind, parallel (3 arm), placebo-controlled trial of 26 weeks duration.</TranslatedText>
	</Description>
	<StudyEventRef Mandatory="Yes" StudyEventOID="SE.SCREENING_VISIT"/>
	<StudyEventRef Mandatory="Yes" StudyEventOID="SE.AMB_ECG_VISIT"/>
...

This is mostly a container element.

StudyEvent

The StudyEvent is the archetype for a Encounters or Study Visits.

<StudyEventDef OID="SE.SCREENING_VISIT" Name="Screening Visit (Visit 1)" Repeating="No" Type="Scheduled">
  <Description>
    <TranslatedText xml:lang="en">Screening Visit at day -14</TranslatedText>
  </Description>
  <FormRef FormOID="F.DOV" Mandatory="Yes" OrderNumber="1"/>
  ...
</StudyEventDef>
...

Form

The Form is an archetype for activities within a Visit.

<FormDef OID="F.DM_1" Name="Demographics" Repeating="No">
  <Description>
    <TranslatedText xml:lang="en-US">Definition of the variables needed to support the collection of a subject's date of birth, gender, race and ethnicity.</TranslatedText>
  </Description>
  <ItemGroupRef ItemGroupOID="IG.DM_1" Mandatory="Yes" OrderNumber="1"/>
</FormDef>
<FormDef OID="F.IE" Name="Inclusion / Exclusion Criteria" Repeating="No">
  <Description>
    <TranslatedText xml:lang="en">Inclusion / Exclusion Criteria</TranslatedText>
  </Description>
  <ItemGroupRef ItemGroupOID="IG.IE_IN" OrderNumber="1" Mandatory="Yes"/>
  <ItemGroupRef ItemGroupOID="IG.IE_EX" OrderNumber="2" Mandatory="Yes"/>
</FormDef>

Form designs are often driven by other business considerations that are not necessariy aligned with data domains. As such a single Form may match to multiple resources in the source system.

ItemGroup

ItemGroup could be considered as primarily a stuctural concept; it links fields to the form. Rather than attaching directly to the form, the ItemGroup is used to group fields within the form. In the current iteration it is primarily used to represent both repeating and non-repeating sets of fields in a CRF.

An example of this would be a series of assessments done around a dosing event (eg Vitals Pre-dose, 1h Post-Dose, 2h Post-Dose, etc); the observations are repeated using an ItemGroup with the Repeating attribute set to Yes

<ItemGroupDef OID="IG.DM_1" Name="Demographics" SASDatasetName="DMSUBJ" Repeating="No">
  <ItemRef ItemOID="I.BRTHDAT" Mandatory="No" OrderNumber="1"/>
  <ItemRef ItemOID="I.SEX" Mandatory="Yes" OrderNumber="2"/>
  <ItemRef ItemOID="I.ETHNIC" Mandatory="Yes" OrderNumber="3"/>
  <ItemRef ItemOID="I.RACE" Mandatory="Yes" OrderNumber="4"/>
</ItemGroupDef>

Item

The Item represents an individual data collection element on a Form.

<ItemDef OID="I.BRTHDAT" SASFieldName="BRTHDAT" Name="Birth Date" DataType="date">
  <Description>
    <TranslatedText xml:lang="en-US">A subject's date of birth. The complete Date of Birth is made from the temporal components of Birth Year, Birth Month, and Birth Day.</TranslatedText>
  </Description>
  <Question>
    <TranslatedText xml:lang="en-US">What is the subject's date of birth?</TranslatedText>
  </Question>
</ItemDef>
...
<ItemDef OID="I.SEX" SASFieldName="SEX" Name="Sex" DataType="text" Origin="CRF" Length="50">
  <Description>
    <TranslatedText xml:lang="en-US">Sex captures a subject's or non-subject's chromosomal characteristic.</TranslatedText>
  </Description>
  <Question>
    <TranslatedText xml:lang="en-US">What is the sex of the subject?</TranslatedText>
  </Question>
  <CodeListRef CodeListOID="CL.SEX"/>
</ItemDef>
...
<ItemDef OID="I.SUDOSE_BEER" DataType="float" Origin="CRF" Length="6" Name="Beer Amount" SASFieldName="SUDOSBER" SignificantDigits="2">
  <Description>
    <TranslatedText xml:lang="en-US">Beer consumption amount.</TranslatedText>
  </Description>
  <Question>
    <TranslatedText xml:lang="en-US">What was the amount of beer consumed?</TranslatedText>
  </Question>
</ItemDef>

As can be seen from the examples, the attributes on the Item element are aimed at the characterisation of the datapoints on the form - in traditional data capture the data is constrained according to the attributes, e.g. the field will be restricted to values matching the declared DataType and Length. Data transfers to these fields will need to conform.

The ItemDef can have two data related child elements; the CodeListRef (zero or one) and MeasurementUnitRef (zero or more). These are links to CodeList and MeasurementUnit elements.

CodeList

A CodeList is used to constrain the permitted values for a field. Each CodeListItem represents a permissible value for the field; these have a CodedValue and a Decode: the CodedValue is the value intended to be written to the backing store and the Decode is the representation of the CodedValue that is shown to the user.

<CodeList OID="CL.SEX" Name="Sex" DataType="text">
  <CodeListItem CodedValue="F" OrderNumber="1">
    <Decode>
      <TranslatedText xml:lang="en-US">Female</TranslatedText>
    </Decode>
  </CodeListItem>
  <CodeListItem CodedValue="M" OrderNumber="2">
    <Decode>
      <TranslatedText xml:lang="en-US">Male</TranslatedText>
    </Decode>
  </CodeListItem>
</CodeList>

MeasurementUnit

The MeasurementUnit is used to represent the possible units for a physical quantity with a Unit; the Unit can be reused across many elements.

<MeasurementUnit OID="mmHg" Name="mmHg">
  <Symbol>
    <TranslatedText xml:lang="en">mmHg</TranslatedText>
    <TranslatedText xml:lang="ja">mmHg</TranslatedText>
  </Symbol>
</MeasurementUnit>

The Symbol is used for the representation of the Unit; the OID is used as the persisted value.

Linking between ODM Elements and FHIR Resources

As part of the current iteration of the IG the following ODM/XML to FHIR Resource high level concept alignments have been tested for defined activities:

Align the Protocol element with PlanDefinition
Align the StudyEvent element with PlanDefinition (with the kind driving the implementation towards an Encounter )
Align the FormDef with ActivityDefinition

In modelling the H2Q-MC-LZZT study ODM the group used a pattern for the Identifier that was used to link the OIDs from the ODM elements to the FHIR resources. Identifiers can be used to find resources using the common resource search patterns - this will be valuable for reconciling study elements. Shared labelling between the model platforms can be built.

An example is shown here:

{
  "resourceType": "PlanDefinition",
  "id": "H2Q-MC-LZZT-Study-Visit-1",
  "identifier": [
    {
      "value": "H2Q-MC-LZZT-Study-Visit-1",
      "type": {
        "coding": [
          {
            "code": "PLAC"
          }
        ]
      },
      "use": "usual"
    },
    {
      "value": "SE.SCREENING_VISIT",
      "system": "http://www.cdisc.org/ns/odm/v1.3/StudyDef#",
      "type": {
        "coding": [
          {
            "system": "http://www.cdisc.org/ns/odm/v1.3#",
            "display": "OID"
          }
        ],
        "text": "OID"
      },
      "use": "secondary"
    }
}

A sample ODM has a StudyEventDef with an OID SE.SCREENING_VISIT; the PlanDefinition has an identifier that links the planned event in the FHIR design back to the definition in the ODM. This resource could then be requested using the following URL /PlanDefinition?identifier=http://www.cdisc.org/ns/odm/v1.3/StudyDef|SE.SCREENING_VISIT