Structured Data Capture
4.0.0 - STU 4 International flag

Structured Data Capture, published by HL7 International / FHIR Infrastructure. This guide is not an authorized publication; it is the continuous build for version 4.0.0 built by the FHIR (HL7® FHIR® Standard) CI Build. This version is based on the current content of https://github.com/HL7/sdc/ and changes regularly. See the Directory of published versions

Adaptive Forms

Page standards status: Trial-use

Adaptive Questionnaire Background

Adaptive Forms follow an architecture where a questionnaire is not pre-defined, rather the questionnaire is dynamically developed based on previous answers to questions chosen from an item bank of questions. This type of administration is also known as Computerized Adaptive Testing (CAT), or administration of Adaptive Questionnaires. In these cases, a set of questions (typically 4 to 12) are administered from a question bank containing numerous questions. The selected questions are based on Item Response Theory (IRT) algorithms which look at the answers provided and select the next question based on IRT algorithms. The questions are selected to quickly obtain the necessary outcomes (e.g. assessing a person's physical function) using fewer questions rather than administering a large question bank. These small sets of questions still provide the necessary confidence and scores required to interpret the data appropriately as if the whole questionnaire was administered. he adaptive process relies on the $next-question operation to dynamically determine and retrieve subsequent questions based on previous answers.

Adaptive Questionnaire Administration Abstract Model and Actors

Profiles:
Relevant Extensions:
Example Questionnaires:

The section explains the process of administering an Adaptive Questionnaire and the actors (systems) involved. Figure 1 below shows the abstract model of the administration process.

Adaptive Questionnaire Administration Abstract Model

Figure 1: Adaptive Questionnaire Administration Abstract Model

Actors, Definitions and Data Flow

External Assessment Center: The External IRT based Assessment Center is a system that can administer a questionnaire based on IRT algorithms. The data that is necessary for the administration is only the initial item bank. The External Assessment Center does not need to know about the specific patient identity or any other clinical information. The Assessment Center will use algorithms recommended by the Questionnaire designer to determine how to administer the questionnaire.

Patient Facing Administration App: The Patient Facing Administration App is the actual app that is being used to present the questions to the patient. It can be an EHR, a SMART on FHIR App, an Independent PRO Administration App.

Data Flow Description for Adaptive Questionnaires Administration

As shown in Figure 1 above, the Patient Facing Administration App (EHR, Care Delivery System, Independent App or SMART on FHIR App) acts as the client and initiates the administration process.

The following section describes the steps that take place when completing an adaptive Questionnaire and also provides examples of the content sent to and received back from the next question operation. With regard to the server's response, it is possible that more than one new question is returned in the questionnaire, and previously sent questions that were unanswered might not be sent back by the server.

Adaptive Population

Adaptive questionnaires can support population under certain constraints. The following guidance applies when using population mechanisms with adaptive forms:

  • Can use observation-based or expression-based, but not StructureMap.
  • Systems should cache previously populated variable values between calls to $next-question.
  • New variables and even libraries may appear in subsequent iterations of the Questionnaire and clients will need to look for these.

Security Considerations

Because the adaptive process involves sending answers to a separate endpoint in order to determine the "next question", the connection should be appropriately protected based on the type of information that might be exchanged. (And given that there is no way to be completely certain what questions will be asked, it is best to err on the side of safety and presume that sensitive information might be shared.) As well, there should be trust between the Form Filler and the nextQuestion endpoint as the server performing the operation will obviously have access to that information. In general, servers performing the nextQuestion operation SHOULD NOT take any action with the data other than to return the response and potentially log the information. "Submission" of a completed form should always be a distinct action taken by the client.