My apologies for being so thick and not directly answering the original question!

Page 16 of the Implementation Guide, Section 3.1.8, provides an example of why the <id> tag has utility in the HL7 SPL document.

As to why it is required in every section, that would be a standard requirement for external accessibility, I think. Perhaps, that the tooling does not use the same value for both the section id attribute and the id tag is distracting; there is no reason that they could not be the same except that you could use a "friendlier" (and, potentially, not globally unique) identifier for the internal reference.

At the end of the day, you're going to need to follow my original advice and ask what the purpose of the section.id attribute is for to HL7's BR&R workgroup or to the FDA. There is nothing in the Implementation Guide that even hints at its purpose. As I mentioned, the equivalent structure in CDA has an optional section.id and I can't recall ever seeing anyone giving an id to a section.

Essentially, the section.id attribute <id root="a857689e-9563-43c0-a244-8a6d5a25966a"/> must be globally unique and assigns a business identifier to the section and can be used externally to the document to refer to that section in a non-technology-specific manner. The section.@ID attribute <section ID=”Section_5.1”> only has to be unique within the document and allows for XML-specific linking of this section to other parts of the same document. So they are serving two different and distinct purposes.

1) If all you had was the XML section.@ID, you could not uniquely refer to that section from outside the document.
2) If all you had was the business section.id, you could not link to the section from within the document.

I think we can all see the benefit of document-linking and thus can see why you would use the section.@ID. What none of us can provide is the rationale for the section.id attribute because I (personally) can't think of a case where you need to have a globally unique identifier for a specific section of a SPL document.

Thank you Jean and Peter, your explanations are good but alas it comes down to (in my mind there are 2 ID's that booth seem to provide the same capability, at least in the envisioned use case).

A snippet from a SPL based Product Monograph (also referred to as a SPM)

<component>
<section ID="L16a947eb-e2be-45c0-8b2e-15d0d0eebed8">
<id root="e6bb83b9-2602-4f96-9077-b8b9535c254e"/>
<code code="30" codeSystem="2.16.840.1.113883.2.20.6.8" displayName="Part II: Scientific Information"/>
<title>Part II: Scientific Information</title>

In this use case the linkHtml would be to the parent element ID (i.e. <section> as the GUID is unique and persistent, but yes we will end up with a id element whether it is <id/> or <id root="e6bb83b9-2602-4f96-9077-b8b9535c254e"/> is now up for debate (your thoughts).

Again thank you guys for your assistance, my main reference for SPL is the FDA's IG but it's silent on this and many, many other aspects.

I think that you explained it correctly, but having an attribute and a tag with the same name is confusing the explanation a bit.

From Jean's example, <section ID=”Section_5.1”> is the "id" that is used for HTML linking, for example, and is completely optional, whilst <id root="a857689e-9563-43c0-a244-8a6d5a25966a"/> is the "id" that is used to provide global uniqueness and is required. The two "ids" serve completely different purposes; the original purpose of the id that is a tag was to indicate the source and, via the extension, to ensure that it was unique.

Is this what was confusing you, Peter Bomberg? Am I totally messing up this explanation, Jean?

Peter,

You can still use a root/extension in the id attribute. But I suspect that, since no business reason is given for the attribute in the first place, that using a GUID instead of a root-OID+extension makes just as much sense for guaranteeing uniqueness.

Like the OP, I am struggling to find a reason for section having a mandatory iD. In CDA, you can give a section an id, but it is 0..1 and I don't recall seeing any in any real-world documents.

Jean

I am sure that someone with more hands-on experience than I have (which includes the OP and JD, for sure ) can comment, but the purpose of the <id> tag is to ensure "global uniqueness," which used to be done by presenting an OID and "extension" used in the local context of the message. With the now-popular tooling, a randomly generated globally unique identifier has replaced the human-readable OID, resulting in the apparent loss of meaning.

See this document by our own Paul Knapp or this comparison of HL7 v2 and v3 messaging for examples of the <id>-as-OID coding.

There might be a good reason to have moved away from using the OID in the <id>, but I will not add speculation to my attempt at explanation.