Wednesday, June 18, 2008

If I had a Hammer

“If the only tool you have is a hammer, you tend to see every problem as a nail.” — Abraham Maslow (1908-70), American psychologist

Recently, I ran across a question on the ANSI/HITSP C32 specification from an implementer. The C32 specification describes how the US Federal Government expects to use the HL7 Continuity of Care Document in our National Health Information Network.

The basic question we started with is how does one represent an operative report, a discharge summary, or a progress note in a C32 document.

I am still a bit confused with the 2 sections and what goes into each one, and the fact that we can't capture the document type (i.e., operative report vs. discharge summary vs. progress note).

The question expresses two parts of the problem:
  1. How do we make use of the CCD to create a uniform way to exchange information, and having done so,
  2. How do we classify documents as to the type of service they describe.
Fortunately, during the development of the CCD, the HL7 Structured Documents work group realized that the CCD templates for problems, medications, allergies, et cetera, could be reused in any other document built on the HL7 CDA standard. Realizing that, they created template identifiers that would allow these templates to be used in any kind of document.

Much work has been done on creating CDA documents using these CCD templates, in HL7, IHE and elsewhere. To date, there are implementations guides that have been developed by IHE and HL7 for:

Each of these implementation guides is using templates from the HL7 Continuity of Care Document, so that as we look inside each document, problems, medications, allergies and other clinical information have a uniform representation based on the CCD.

So the real question is not, "how to I put an operative note into a CCD", but rather,

How do we use the CCD specifications to record this information inside a _____(fill in the blank)?

The answer is to use the same templates created for the CCD, and in the ANSI/HITSP C32 specification in those other clinical documents.

The reason for doing this can be explained very simply. Imagine that you are a physician caring for a patient, and you want to find a particular document. It may be an operative note, a discharge summary, a consultation, et cetera. If all documents are CCD's, then you no longer have the capability to distinguish between them by the kind of service represented in the document. Can you imaging wading through all of these CCD's to find the right document? What you really want is for those documents to be classified by the kind of service performed by the provider. If it's a discharge summary, then it should say so.

Recently, the ANSI/HITSP Care Management and Health Records Technical Committee met in Washington DC, and discussed this topic. The solution proposed was to recognize that the C32 specifications applied not to just CCD documents, but to all HITSP created CDA based specifications. That committee will be reworking the HITSP specifications this year to better enable reuse of the C32 specifications across all clinical documents.

The HL7 Structured Documents work group will also be reviewing plans for development of the next release of the HL7 Continuity of Care Document later this year. The expectation is that we would propose changes based on feedback from implementers such as the NHIN implementation projects. When that work begins, I will be proposing a change to the very first conformance statement in the Continuity of Care Document. That statement is reproduced here:

CONF-1: The value for “ClinicalDocument / code” SHALL be “34133-9” “Summarization of episode note” 2.16.840.1.113883.6.1 LOINC STATIC.

My proposal will change this conformance statement to allow for the use of other LOINC codes when the purpose of the document is to contain a medical summary and documentation of other care. I've shown the proposed text below. Please note, this is only my proposal, there is no guarantee that this will become part of the next version of CCD, but it could help to address the confusion raised by the current specification.

CONF-1: A document conforming the these specifications may use any LOINC code in “ClinicalDocument / code” to describe its content. When the purpose of the document is solely to summarize the patient's current health status, the value for “ClinicalDocument / code” SHALL be “34133-9” “Summarization of episode note” 2.16.840.1.113883.6.1 LOINC STATIC.

We need use the right tool for the job, and while the CCD Document is a hammer, the CCD specification itself provides us with a whole box of tools -- CDA sections and entries. We can use these tools to build any number of clinical documents. Furthermore, having done so, we can expect that healthcare applications will be able to understand the content.


  1. Part of this is a branding and market education exercise - beyond the technical aspects. In HITSP I hear people use "CCD" meaning "a clinical document" rather than "the compromise developed between HL7 CDA and ASTM CCR". What we need to do is use CDA to mean "clinical document", and correct those who use "CCD".

    What is being done in IHE and HITSP is a more direct use of CDA and reuse of templates -- something that ASTM did not include in the CCR standard.

  2. It's also worthwhile (for non-US readers) to point out that the CCD (as an overall CDA document) is US-specific. However: theses specifications are still of value, because they define templates which can also be used outside of the US.
    The CDA documents created by IHE PCC (and some others) re-use some of the CCD section/entry level templates. An IHE PCC CDA document, when implemented outside of the US, does in no way have to comply with CCD. Different countries have diferent requirements ..

  3. Actually Ren, I have to differ with you a little on the statement that "An IHE PCC document... does in no way have to comply with CCD." The IHE PCC CDA templates do use some of the CCD templates, and so some compliance is necessary. However, to your point, the CCD templates were built based on best practices from the various HL7 domain committee models, and so are completely usable in an international setting. We've already see several countries doing so, and I've seen some implementation guides in Europe making use of the templates based on CCD that are found in the IHE PCC Technical Framework.

  4. May I propose two tweaks to your proposal to the HL7 Structured Documents committee?

    1. instead of "any LOINC code", "any LOINC Document code". Unfortunately, LOINC is a bit sloppy in how you can identify a code as being a document code, but I would propose "any LOINC code with a scale type of 'Doc'", and hope that we can convince LOINC to clean up these codes so that this is accurate.

    2. Instead of "When the purpose of the document is solely to summarize the patient's current health status", how about "when the purpose of the document is to document a clinical episode or encounter". I think this more accurately captures the nature of the CCD spec and the 34133-9 LOINC code. Some "episode summaries" may well include everything about "the patient's current health status," but we shouldn't expect or want all such documents to. In an ideal world, there might be one active "current health status" document, authored by the patient's primary care provider (or maybe by the patient in a PHR), but subsequent encounters with specialists would have no need to reproduce all that information that is outside the subject of that encounter.