On the last SHR community call I offered to post up some information regarding CDA content specifications that may be used to develop underlying database schema for our SHR. I found a “matrix” of the various data elements (HL7 templates) and the CDA “sections” in which those elements appear. This matrix can be found here: http://cdatools.org/ClinicalStatementMatrix.html.
To see how these sections and elements (all of which are curated and identified with HL7 OIDs) may be collected together into a clinical document, the IHE CDA specification for discharge summaries is a good example. The two part IHE technical framework for patient care coordination (PCC-TF) can be found here:
http://www.ihe.net/uploadedFiles/Documents/PCC/IHE_PCC_TF_Vol1.pdf (integration profiles)
http://www.ihe.net/uploadedFiles/Documents/PCC/IHE_PCC_TF_Vol2.pdf (content profiles)
Note in volume 2, pages 34-35 explicitly lay out the content templates that are in specific document types. On pages 48-50, the lists of sections for a discharge are listed. The expanded “elements” definition for each section is expanded upon in its respective section (including data typing and code set specifications, as applicable). In total, this second volume gives examples of how the structure and content of a discharge is described. The detail of this specification could be used to map out a database structure (or could be employed, via an XML database, to natively navigate to specific data elements). There is a lot of experience within our friends and Mohawk regarding how to automagically parse content from HL7v3 constructs (CDA employs HL7v3 models).
I hope this information is helpful.