Representation of the structured sections in the composition

[gcangioli]

The HDR model includes some very strcutured sections see e.g http://hl7.eu/fhir/hdr/StructureDefinition/AdmissionEvaluation or HospitalStay or the recommendations section. that includes also the "medication recommended for the period after discharge".

A similar structured structure was present aso in the PS model that have been at the end represented as a flat slit of content-oriented sections.

Should the Composition structure reflects the model structure or it should be flat as it is for the PS and substantially in the CCDA Discharge Summary https://build.fhir.org/ig/HL7/ccda-on-fhir/StructureDefinition-Discharge-Summary.html

[gcangioli]

The current proposed approach is highly structured (maybe even more than needed) but in the description of some sections it seems some informaiton as medications or problems are captured in a comon single section..

[mgraauw]

If we go for structured Composition as in the Logical Model there are some cross-issues:

HDR has:

• A.2.6.1 - Medical history
• A.2.6.1.1 - History of problems
• A.2.6.1.2 - Devices and Implants
• A.2.6.1.3 - History of procedures
• A.2.6.1.4 - Vaccination

In PS there is:

• A.2.2 Medical history
• A.2.2.1 Vaccination/ prophylaxis information
• A.2.2.2 Resolved, closed or inactive problems
• A.2.2.3 Medical history

Also w/o devices and procedures which are in:

• A.2.3 Medical problems
• A.2.3.1 Current problems
• A.2.3.2 Medical devices and implants
• A.2.3.3 Procedures
• A.2.3.4 Functional status

So the meaning of "Medical problems" as a section is different in HDR and PS, and there are 2 "Medical history" sections in PS, one a composite one (A 2.2) and a narrative one (A.2.2.3).

If we flatten the PS and HDR in a flattened Composition with just the lowest section level:

• is is much easier to have - where the content fits - the same section in HDR and PS
• we avoid the problem of having to find different LOINC codes for "Medical history" (3 times), as well as some other incongruencies between PS and HDR Logical Model

[mgraauw]

Capturing (I hope rightly) the discussion on 22 Nov 2024: we start with a "library" of low level sections and for now a flattened Composition. On al logical model (EU, national, or other) those can be aggregated in composite sections and a mapping can reflect that. Open question is whether there is an appropriate code for all such sections which captures "staging" semantics (admission, stay, discharge) but we can see that when we model the low level sections.

[mgraauw]

And there was the comment that the incongruencies between HDR and PS logical models I mentioned above are being addressed within XtEHR

[rlindstrm]

Question from Kari: I might be missing something obvious but the HDR Composition section:sectionAdmissionEvaluation (LOINC 67852-4) and section:sectionDischargeDetails (LOINC 8650-4) seem to have similar sub-sections as sliced by SCT codes i.e. currently the hierachy is used to nominate the context. How would these work in the flattened model ? Find new 'magic values' that combine current stage designation (LOINC) and data category (SCT) ?

[gcangioli]

Decision from the 2024-11-22 meeting "We offer library of sections in a flat structure, and provide mapping to models. Implementers are allowed to use subsections."