This page is part of the IHE SDC/eCC on FHIR (v1.0.0: STU1) based on FHIR (HL7® FHIR® Standard) 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
IHE Structured Data Capture (SDC) on FHIR uses a form driven workflow to capture and encode data by creating FHIR Observations or other relevant resources from the captured data. This guide will focus on 2 types of data extraction: Observation-based and Definition-based. These concepts are borrowed from, and aligned with the HL7 FHIR SDC Implementation Guide: Form Data Extraction guidance.
Observations can be used to capture Question and Answer Sets from SDC. Each Observation should capture the Question ID from SDC in the code element, and the answer in the value element. There are several question types that are described below on best practices for capture. The following sections will describe how to generally map some common data elements from IHE SDC to FHIR Observations, then will describe special considerations for specific data types.
Questions in IHE SDC are the element which are displayed to the user for them to fill in. There are a variety of question types including ListItems, Booleans, Strings, integers, decimals, etc. This page will detail how to map all of those question types to FHIR Observations.
Since IHE SDC is XML only the different elements and attributes will be identified from that specification below. Elements will be identified as starting with a capital letter (e.g. Question
), while attributes will begin with an @ (e.g. @title
). Attributes belong to one or more elements, so if a specific attribute must be used its parent element will be called out. IHE SDC elements and attributes will be described in XPath, such as Element/Element/@attribute
. W3C offers a XPath Tutorial. FHIR elements will be described in the standard FHIR Notation of ResourceName.dataElement
. For example, we map the IHE SDC attribute Question/@instanceGUID
to Observation.identifier
.
The IHE SDC/eCP on FHIR Observation profile is designed to be used with all the following data types in the subsections below.
The IHE SDC/eCP on FHIR Observation profile requires Observation.identifier to be used and supported. Observations should be identified using the IHE SDC Question/@instanceGUID
to the Observation.identifier
. The @instanceGUID will uniquely identify this Observation. Observations should reference a patient, the practitioner, a date, and be attached to a DiagnosticReport to ensure provenance of the resource.
Said DiagnosticReport should conform to the The IHE SDC/eCP on FHIR DiagnosticReport profile.
For all Questions in IHE SDC the @ID and rptTxt or @title should be captured in the Observation.code
element.
The text to include in the Observation.code
should come from one of the following IHE SDC elements or attributes:
Property/@propname="rptTxt"
@title
rptTxt
or @title
may be found as part of an IHE SDC Question or ListItem. rptTxt will not be present on every IHE SDC Question or ListItem, but should always be used in place of @title if available.
The Observation.code
element should be used to represent the @ID
and @title
of IHE SDC Question, regardless of question type.
For guidance on how to add standard terminology to these Observation.code
elements, see the Terminology section of this IG.
XPath: Question/Property/@propName="reportText"/@val
For Single and Multi-select questions the implementer must use the report text (rptText) attribute instead of the title attribute which is the default. rptText should be mapped for FHIR Observation.ValueCodeableConcept.
There may be some reportText properties which have a @val={no text}
. In these cases the @title attribute should still be used.
XPath: Question/@title
If Report Text has is not present then the report @title
of the answer SHALL be used as the ValueCodeableConcept.
As Observations represent the Question and Answer pairs from IHE SDC, they may have parent or child questions, which must also be presented in Observations. To preserve the hierachy of Question and Answer pairs the Observation.derivedFrom
element should be used to capture the Observation.identifier
of the parent question and Observation.hasMember
should be used to capture the Observation.identifier
of the child question. The Observation specification provides guidance on how to use Observation.hasMember and Observation.derivedFrom. A parent question may have multiple children within its Observation.hasMember
field.
An SDC eCP Observation Parent Example and SDC eCP Observation Child Example showing how these Observations can be grouped together can be found in the Artifacts page.
For all Sections and Questions, they only reference child and/or parent Questions/Sections. Never a Grandparent, Grandchild, etc.
Observations groupers should be used to represent sections as a parent Observation. When representing a section in an Observation the implementer is expected to leave the Observation.value[x]
element blank. The Observation.code
element should represent the Section’s @ID
and @title
in IHE SDC.
All sub-questions of the Section should be captured within the Observation.hasMember
element.
SDC eCP Observation Section Example
IHE SDC has two kinds of List items:
For both of these List item types the IHE SDC Question
should map to the Observation.code
, as noted above, and the ListItem
should map to the Observation.valueCodeableConcept
.
Like Questions, all ListItem answers will have an @ID element which should be captured in the Observation.valueCodeableConcept
element along with the rptTxt property or @title attribute.
The same guidance in the Terminology section of this IG for Observation.code
applies to Observation.valueCodeableConcept
.
Single Select question/answer pairs in IHE SDC should be captured as Observation.valueCodeableConcept
with the IHE SDC ListItem @ID
for as the coding, and the ListItem
@altText, @rptTxt, or @title as text.
Similar to Single Select question/answer pairs in SDC should be captured as Observation.valueCodeableConcept
with the SDC ID as the coding as separate Observations. Therefore each answer should be uniquely represented in a separate Observation and could be linked together by referencing the same DocumentReference in the Observation.derivedFrom
element or in a FHIR DiagnosticReport.
There are a number of questions in IHE SDC which have sub-question types directly attached in the parent. These are known as ListItemResponses
. For example a single select list may have a question like “Tumor Size” with answer options of:
For “Greatest dimension in cm” there may be a numerical fill in box attached the question, and likewise for the “Cannot be determined (explain)” there may be a string List Item Response box attached. These are handled as special elements from other questions in the IHE SDC specification known as ListItemResponseFields
and require the Observation to the component element to capture the additional filled information. See example: SDCeCPObservationCodeLIR
Observations for text based questions should be captured as Observation.valueString
for simplicity.
DisplayItems are text which are shown to the user filling out the form. These elements do not need to be represented within FHIR.
There are some questions which may not have an @title
attribute or a rptText element present in their question element. These untitled questions are typically sub-questions to be answered in specific contexts. If an untitled question is present then the untitled question @ID
should be included as the Observation.code
element, with a blank display value.
Other data types should be captured in Observation.value[x]
as their closest IHE SDC for match, whether that be DateTime, Boolean, Integer, etc.
IHE SDC DataType | FHIR Observation.value[x] type |
---|---|
boolean | valueBoolean |
date | valueDateTime |
dateTime | valueDateTime |
decimal | valueQuantity |
integer | valueInteger |
string | valueString |
time | valueTime |
The Structure Map may be used to cover complex types of questions which have unique needs in terms of mapping to FHIR resources. For example if a group of questions needs to be represented in a single Observation a Structure Map could specify how to add those answers to Observation.component elements. This would enable more complex and accurate mapping to FHIR resources.
Additionally Structure Map could be used to write to different FHIR resources other than Observation, where a resource such as Condition or Procedure may be more appropriate.
Note: Structure Map is still being explored by the IG authors for these kinds of use cases.
It is possible that the patterns detailed in this guide could be used to extract data from historically collected IHE SDC forms. If the question types and IHE SDC IDs were maintained it could be feasible to write that data into FHIR messages following profiles outlined in this guide.