Page tree
Skip to end of metadata
Go to start of metadata



Owning work group name:

Clinical Decision Support



Committee Approval Date:

Initial PSS: June 22, 2018

Resource proposal approved  2020-10-21 Meeting Agenda

Note this is not an approval of the resource as maturity level 1, that approval is still outstanding


Contributing or Reviewing Work Groups:

Clinical Decision Support, Biomedical Research and Regulation, Clinical Quality Information


FHIR Resource Development Project Insight ID:

1422


Scope of coverage:

The scope of EvidenceReport is any composition which is about community evidence (ie not patient-specific) and may include any combination of citations, evidence (results), evidence variables, and classifiers.



RIM scope:

out of scope for the RIM


Resource appropriateness:

Across the academic and scientific community there is an extensive need to report evidence in a nearly infinite variety of combinations. In addition to the structure of the "output" or "results" of the evidence report there is a need to classify what reports are about, and to support precise searching for evidence in many different combinations. Standardization of expression of the focus or subject of search queries, search results, and specific artifacts (notably EvidenceReport compositions) is needed to increase efficiency and precision across multiple groups accelerating the uptake of computable expression of biomedical evidence.

FHIR Management Group questioned if this could be done as a Profile of the Composition Resource.   We mapped out 8 example reasons to use an EvidenceReport Resource including:

  1. List of citations related to an evidence concept (eg. PICO question)
  2. Summary of Findings table from a systematic review
  3. Summary of all study results for a single PICO question
  4. Organizing search results for queries specifying evidence/evidence variable/citation concepts
  5. Collection of “all the data” (Citations, Evidences (study results, meta-analysis results, statistics and certainty), EvidenceVariables (PICO coding), Narratives (plain language summaries)
  6. Creating alerting reports by specific search queries
  7. Summary of all or selected evidence related to a single study report
  8. Methods to search, export and view any of the evidence/evidence variable/citation concepts


Comparing the Composition Resource to the EvidenceReport Resource shows the following differences which explains the challenges with using Composition Resource for evidence reports:

  1. Composition.status uses the codes preliminary/final/amended/entered-in-error and EvidenceReport.status uses the codes draft/active/retried/unknown – the PublicationStatus value set is more relevant for evidence reports than the CompositionStatus value set. The Evidence Report is more like a metadata resource pattern than a patient-specific communication document.
  2. Composition does not have a citeAs element. This is uncommonly needed for patient-related compositions but frequently needed for evidence reports.
  3. Composition.type is required (1..1) – we relaxed this in EvidenceReport.type (0..1) because this is not necessary for function.
  4. Composition.subject (0..1) requires a single resource or single text string to describe the subject.  EvidenceReport.subject (1..1) is a BackboneElement supporting 0..* characteristic elements and 0..* note elements to describe the subject which is much more flexible for the many uses of EvidenceReport.
  5. Composition.encounter was removed in EvidenceReport as it is not about patient encounters.
  6. Composition.date (1..1) is a single element for “Composition editing time”.   EvidenceReport is a metadata resource pattern so inherits date, approvalDate and lastReviewDate elements --- more date elements are needed for evidence reports than for compositions.
  7. Composition.title is required. It is for a human-readable name/title, and there is no element for a machine-readable name.  EvidenceReport inherits name and title elements from the MetadataResource pattern, and there are use cases that support machine-only and human-only names.
  8. Composition.attester requires a mode element which requires a choice of personal, professional, legal or official. EvidenceReport does not require such legally binding concepts and provides optional elements for reviewer, endorser, and editor roles.
  9. Composition.relatesTo.code requires 1 of 4 choices (replaces, transforms, signs, appends). EvidenceReport.relatesTo.code requires 1 of 8 choices.  ‘signs’ is not relevant. ‘amends’ was added. Also all 4 codes have the opposite-direction codes for bidirectional relationships.
  10. Composition.event was ignored for EvidenceReport which does not related to “The clinical service being documented”
  11. Composition.section.entry (limited to Reference datatype) was expanded to support entryClassifier and entryQuantity – the entryClassifier with datatype CodeableConcept is highly used in example EvidenceReport instances.

We generated 18 EvidenceReport Resource examples (found at https://fevir.net/content/resources with titles starting with COKA-EBM) to show different uses of EvidenceReport. For example:




Expected implementations:

Immediate implementation is expected with the COVID-19 Knowledge Accelerator project and initial implementers include Computable Publishing LLC, The MITRE Corporation (for the AHRQ CEDAR project), and PICO Portal.  Fast followers are expected to include Agency for Healthcare Research and Quality (AHRQ) Systematic Review Data Repository (SRDR), Cochrane, Duodecim Medical Publications Ltd (from the Finnish Medical Society), MAGIC (a project in Norway), and multiple consortia coordinating evidence processing for COVID-19 such as COVID-END.



Content sources:

No specific external content sources


Example Scenarios:

A large number of groups are classifying COVID-19 articles for multiple reasons: Do they include research results for human COVID-19? Are they about treatment, prevention or prediction? Do the methods include randomized trials or uncontrolled observations? These groups need to share a common EvidenceReport (in this case the classifiers related to a Citation Resource for the article being classified) to coordinate independent classification efforts.

A clinical decision support vendor wants to report the collection of evidence results for benefits and harms of a specific intervention.

A researcher presents a search query for evidence about a specific population and meeting selected characteristics (such as study type randomized trial) and the search results are compiled using an EvidenceReport Resource.


Resource Relationships:

The EvidenceReport Resource may reference Citation Resource, EvidenceVariable Resource, EvidenceReport Resource or any other FHIR Resource.

The EvidenceReport Resource may be referenced by the EvidenceReport Resource.


Resource Boundaries:

No areas of confusion discovered yet.

Timelines:

Pre-applied (http://build.fhir.org/evidencefocus.html) for Connectathon 24.   Approval (at Maturity Level 0) was applied for R4B.


gForge Users:

brynrhodes (github user) Khalid-Shahin (github user)


When Resource Proposal Is Complete:


When you have completed your proposal, please send an email to FMGcontact@HL7.org

FMG Notes