Monday, March 26, 2007

Interesting Post from HL7 Connection re HL7 V3

From the HL7 Connection blog:

"Has anyone successfully implemented version 3? In my spare time I have been reading everything I can find on best practices and potential pitfalls."
Posted 3 days ago #

"Why? There is no viable business model. The vendors with current market share have been quite sucessful with v2 interfaces. There are no clear advantages. The v3 data model is not a model of the interface record but rather of an application database. Why would vendors abandon their proprietary advantages developed at significant expense in order to re-develop their product as a comomdity offering, particularly when there is limited, if any, consumer demand (at least in the US).

The only US vendor (if you can call them that since they have no application product that has been implemented in a sucessful domestic healthcare setting) is Oracle. Last time I looked their only sign-up was Quadramed - which has since abondoned the effort (or so I was told by one of their application develoment managers.

Right now, if I were starting fresh, I'd implement v2 using XML.

While I'm at it, where is the formal transition strategy for mixed v2 & v3 interface settings. In my thirty some years in the field I've learned that the transitions are a very big deal.
This issue has popped up at the technical steering committee meetings for well over half a decade with no progress seen, or am I wrong?

What made HL7 such a success was its proclivity for the practicle real world needs of healthcare organizations and vendors. Now we have the Europeans mucking up HL7 the way they did CEN TC251.

The English NHS project is several years behind schedule, over budget, and gaining a reputation for 'shooting the sled dog vendors in the head' (thats a direct quote from a the senior english Government offical) - and they have taken v3 and bastardized it to the point that it could not be considered a standard approach to application interfaces outside the NHS.

The proof comes when one learns that the HL7 Board tried several years ago to discontinue further iterative enhancements to v2 telling committee chairs (I was one)that they should abandon additional v2 development. This edict was later recinded. I would see this as a clear indication that it was the result of the healthcare community having their say.

In my humble opinion, after more than a decade of development, HL7v3 is stillborn. I'm very unhappy about the situation and I very much hope I'm wrong. Please show me the error of my thinking - believe me when I say I'd be delighted. Until then I'll continue to recommend the use of v2 interfaces using XML."

1 comment:

Marcelo said...

Hi, Barry. I work in Medical Informatics in Argentina and I think that the best news about HL7 is that the v2.5.1 is realeased. I recommend to my clients that the best option right now is to use v2.x with XML. With the advance of Webservices I don't see the need for communication protocols and other messaging specifications that in the past was a very useful feature of the HL7 messaging specification.
In my modest opinion,I believed that HL7 must begin to work on version 4 (dropout version 3) and focus seriously on Ontology and not on this hibrid v3 mix of methodology and reference model. The other problem is may be that Ontology will affect other business of HL7 like CDA not only messaging.
Last, I want to say that when I talk about HL7 I talk about the core or most important business of HL7 that is messaging and not about other business like CDA, CCOW or Arden.