Medical Informatics

Back to Medical Informatics this evening ..

a) There is a new report out from the GAO.  It's 120 pages long, but worth the download if you're intersted.  Here's the overview:

The rapidly rising costs of health care, along with an increasing concern for the quality of care and the safety of patients, are driving health care organizations to use information technology (IT) to automate clinical care operations and their associated administrative functions. Among its other functions, IT is now being used for electronic medical records, order management and results reporting, patient care management, and Internet access for patient and provider communications. It also provides automated billing and financial management.

The Ranking Minority Member of the Senate Committee on Health, Education, Labor, and Pensions asked GAO to identify cost savings and other benefits realized by health care organizations that have implemented IT both in providing clinical health care and in the administrative functions associated with health care delivery. GAO analyzed information from 10 private and public health care delivery organizations, 3 health care insurers, and 1 community data network.


b) In my discussions of the AMIA meetings from last week – I didn't discuss two topics that were more important than I expected.

  1. CCR.  I mentioned it briefly last week (and last May) but it deserves more explanation. CCR is an attempt to develop a minimal data set that would provide enough data for a patient to be adequately cared for in a new setting.  Ideally thought of as a "transfer record" – it's become more than that over the last 6 months, and remains a work in progress.

    CCR is controversial.  Many argue that HL7 could do this just fine, and that there is no reason to develop a NEW standard – where augmentation of the existing data transfer standard would suffice.

    But maybe it wouldn't – and maybe we need to agree on what this minimal dataset is before 2015 – as HL7 (the organization) moves rather slowly.

    So with the framework of ASTM and the support of HIMSS and the Massachusetts Medical Society, CCR is building steam.  A measure of this steam is the fact that the AAFP project (and all of the vendors who have signed on with that project) has embraced CCR as a method of sharing data across systems.

    Hmmm … While this is intuitively good, it's also risky at this point for anyone to be certain that this is good, since the definition of CCR remains up in the air.  the drafts I've seen have had weaknesses – depending on the perspective of the reviewer.  The balance betweek keeping it simple (and therefore easy to implement) and keeping it complete (thereby capturing all of the important data) is a tricky one.

    CCR will be defined in XML.  This is no big deal – and is certainly the appropriate way to express it.  What's nice about XML is that the data can be transformed very easily depending on how the user wants to see it.  If you're curious what in the world I'm talking about – you may need a tiny intro to XML  Here you go.  Just read the 1st page — shouldn't take you more than 2 minutes.  I'll wait here until you come back.

    Ok .. so we all "get" XML now.  See .. it's really pretty simple and indeed that is the goal of CCR.  To define a patient record in a simple way so that we can transfer care quickly and easily.

    Microsoft is building new XML tools such as InfoPath.  They understand that healthcare is the right kind of business to market this stuff to.  Here's an interesting demonstration of Infopath:  The Microsoft Office InfoPath 2003 HL7 CDA Demo.

    Should we start interrogating our vendors on whether they support CCR?  No .. not yet .. but if they haven't heard of CCR at all – we may want to give them some homework.  To get more involved in the CCR project and review drafts, give input, etc .. join ASTM.

  2. Along similar lines, there is a confusing project now going on within HL7.  When many people think of HL7, they think of the protocol that defines transactions.  IN fact, HL7 is an organization and within this organization, there are a group of people who are trying to answer a question that the government has charged them with answering:  what is an electronic health record?

    It's and important question – since the answer will ultimately define a set of standards that will, one hopes, provide better uniformity in how these systems function.  Think of medical records in our practices and hospitals like money in banks.  It doesn't really matter to the banks how the money is stored electonically.  I'll bet that the database at Bank of America looks different from the database at Wells Fargo.  But since the banks have all agreed on how the money will go out and come in … my money can move from one bank to another pretty well.  The standard that defines how the money moves is like the HL7 standard for transactions — and in some ways, like the CCR.    My bank account is like my electonic health record.  If we can't agree on what a bank account (electonic health record) is, then it's going to be hard to form the standard for the transactions between the banks (healthcare providers).

    So what?  well .. if one system understand the concept of "family history" and the other one doesn't .. then the system there is no way that this data is going to come across from one system to the other.  So we have to define what the systems will be able to understand.

    This is what the HL7 EHR functional Model  is. 

    I went to a session hosted by the HL7 folks.  It was billed as a "town meeting" but was rather sparsly attended.  For two hours, we argued about what an EHR is, and how it should be described.  I was especially impressed with the insight of Don Nelson.   (scroll down to see Don)