NEHTA Clinical Documents: Conformance Levels and Diagnostic Investigations

Sep 3, 2012

CDA defines 3 conformance levels: 1. The unconstrained CDA specification.

  1. The CDA specification with section-level templates applied.
  2. The CDA specification with entry-level (and optionally section-level) templates applied.

In practice, this is equates to:

  1. A document with just a header and a single narrative
  2. A document with a header, and a narrative broken up into sections (that conform to some template - i.e. implementation guide)
  3. A document with a header, sections, and entries containing data as defined by some template)

These are only loose definitions, as are the formal definitions of the 3 conformance levels - they do not precisely define what a document should have, only it’s general shape.

In NEHTA, we take those levels a step further:

  1. A document with just a header and a single narrative
  2. The narrative is an attachment (i.e. PDF)

  3. The narrative is CDA narrative

  4. A document with a header, and a narrative broken up into sections (that conform to some template - i.e. implementation guide)
  5. A document with a header, sections, and entries containing data as defined by some template)
  6. The coded entries are present
  7. The coded entries are present and coded with the expected coding system (usually Snomed-CT AU or AMT)

Most documents currently being prepared for the pcEHR contain a mix of level 2 and level 3 sections - for instance, they might have a purely text section for medical history on the basis that this is all they are able to retrieve from the responsible sub-system, while containing a fully coded medications section, since the medications sub-system makes that information available. Note that a document can be level 1, or some mix of level 2 and 3 on a per-section basis. Also note that to my knowledge there are no level 3B documents in preparation, though some of the documents I’ve seen include some specific items that are coded in Snomed-CT AU or AMT.

Narrative in Attachments

With regard to level 1 documents, originally it was expected that a level 1A document would be represented using a nonXMLBody element instead of a structured body. However for various reasons the logical header content overflowed into the structured body (a section called “Administrative Observations”), and this can’t be included in the body if you use nonXmlBody. Also, practice shows that most systems that are able to source a PDF document are also able to source some atomic data items which would be good to include in the document, but can’t be if you use nonXmlBody. For this reason, level 1A documents are expected to be represented in the following way:

<component>
 <structuredBody>
  <component>
   <section>
    <id root="..." />
    <code code="..." />
    <title>[Name]</title>
    <text>
     <renderMultiMedia referencedObject="PDF1">
      <caption>[Name] Representation</caption>
     </renderMultiMedia>
    </text>
    <entry>
     <observationMedia ID="PDF1" classCode="OBS" moodCode="EVN">
      <id root="..." />
      <value xsi:type="ED" mediaType="application/pdf">
       <reference value="[url]" />
      </value>
     </observationMedia>
    </entry>
   </section>
  </component>
 </structuredBody>
</component>

In the NEHTA context, the url is a reference into the package containing the CDA document. This looks exactly the same as a nonXmlBody, but allows the document to also contain some selected data items.

Diagnostic Investigation Conformance Levels

Diagnostic Investigations (i.e. Reports) are included in the clinical documents, but as I’ve said before, they are different, because each diagnostic investigation is in effect it’s own document, and it’s being enclosed in a secondary use clinical document. So we could define the same conformance levels for a diagnostic investigation:

  1. Purely a presentation only (i.e. just the report)
  2. The presentation is a PDF

  3. The presentation is CDA narrative

  4. The presentation, but also with information to properly identify it - report type, date, and status
  5. Fully structured report with atomic data
  6. The structured data is present
  7. The structured data is present and coded with LOINC and Snomed-CT AU

These levels relate to each differently. It’s possible - and quite reasonable - to provide both a CDA narrative representation, and a PDF representation. These forms have different strengths and weaknesses, and providing both allows the displaying system to choose which is the better to display (of course, you have to be sure they say the same thing, but this is often possible). In addition, it’s sensible to provide both the presentation and whatever structured data is possible - so to combine levels 1 and 2+.

Note that with regard to structured data, there are many aspects of structured data, and level 2/3 doesn’t differentiate between metadata, request data, atomic data, specimen information, and attached images, so again, the conformance levels are loose.

Because the diagnostic investigations section is assembled from other sources, the content rules in it are different to the rest of the document. The basic structure of the diagnostic investigation, as covered before, allows for:

  • PDF representation
  • CDA narrative representation (if you have both, you put the link to the pdf - using the reference form shown above - at the bottom of the CDA representation)
  • Report type / status / date
  • Atomic data, request data, specimen data, metadata, images

In the CDA IGs for the clinical documents, the following items are required (minimum cardinality 1):

  • Some form of narrative (may be just a reference to an attachment)
  • Type
  • Date
  • Status

The implication of this is that a clinical document that has level 3 conformance is required to have a least a level 2 diagnostic investigation. On the other hand, if a clinical document only claims level 2, then the cardinalities on the structured data items do not apply, and level 1 diagnostic investigations can be included. (btw, the implication of this is that if you can’t represent diagnostic investigations at at least level 2, then you either omit them from the document, or only claim level 2 conformance, even if the rest of the document conforms to level 3).

About claiming conformance levels

Just a note about claiming conformance levels: the clinical documents themselves do not contain any reference to what conformance level is asserted for them. All they contain is a reference to the CDA implementation guide itself, and therefore it may be generally level 1, 2, or 3, and this may vary by section, and it may or may not contain some or all of the data entries that are defined.

The pcEHR, on the other hand, will check that a document is valid when it is submitted; if it’s not valid, it will be rejected. The validation check is against the level claimed about the document when it is submitted to the pcEHR. That claim is made in the XDS metadata (also is found in the metadata when the document is retrieved from the pcEHR). For example, if the document claims level 2 conformance, then it will contain the mandatory sections as defined in the IG - but note that it may also contain additional structured data in some sections (only this won’t have been checked by the validator, so document processing systems should beware when extracting and using this data).

Document authors should be aware that although the pcEHR does not enforce the rules made in the CDA IG about the structured data if the document claims level 2 conformance, the only rules that are actually relaxed are the minimum cardinality - the documents are still required to be otherwise conformant to the CDA IG.