The Swedish vs US approaches: health information, not just communication

National approaches to eHealth program development have been varied, and unfortunately we can all probably identify more failed attempts than successes. Just recently we have seen a spectacular regrouping within the UK NHS in the face of cost blow-outs and underperformance - a stellar example of how a project commonly described as the biggest computing exercise in the world, outside the military, has hit the proverbial brick wall. My question is not so much "What is the 'right' approach?" but more importantly "What is a sustainable approach, so that our valuable health information will persist at least as long as we do, and hopefully as long as our children need it?" Brian Ahier (@ahier) triggered some thoughts when he sent me a link to the new O'Reilly Radar article, Healthcare communication gets an upgrade which he co-authored with Rich Elmore and David C. Kibbe. It is an informative and succinct post that explains the US Direct Project (formerly NHIN Direct) in words that actually make sense to this non-technical clinician! No doubt it is good infrastructural work that will be very valuable.

However, given my day-to-day work with archetypes and standardising clinical content in Europe and Australia, the 'elephant-in-the-room' for me in this article is what is the clinical content that is to be exchanged? I assume we are talking about structured clinical documents as CDA, and similar. And that is great for short- to medium-term gains but, in the longer timeframe, as we want to share more data, and more complex data, and then we also expect to be able to compute upon that data rather than just read it... what then? How do we make sure that the data is safe and exchangeable between disparate systems? What is the long-term plan? Is a structured document enough? I think absolutely not.

So "What is the plan", I asked today of the authors and the Direct Project. I watch with interest for a response. Is there a strategy for content standardisation?

My understanding of the current situation with content development in the US?

  • Stan Huff's team has done fantastic work at Intermountain Health - a very sophisticated approach, but only used within their group;
  • UK's Charlie McKay tells me that there are currently at least 20 different approaches to clinical content development within HL7 alone that he can identify - a rather unfortunate situation for a standards body;
  • Think of all the clinical application developers in the US alone - more than 7000 vendors I'm told. Even if it is only hundreds, imagine the resources poured into developing the same clinical content in different ways for each separate application's proprietary information model, each reinventing the wheel and only interoperable by accident - what a waste of resources;
  • and then a plethora of others, including Tolven etc.

I have not been able to identify a cohesive approach to defining clinical content such that it is safe to be directly exchanged between clinical systems - am I wrong?

By contrast, only last week there was a series of emails on the openEHR Clinical discussion email list. It was offering a number of links to Sweden's recent very public meeting where they invited experts from throughout Europe to share the Swedish national eHealth program and plans, and to receive feedback from their EU colleagues. Taking a totally orthogonal approach, the Swedish priority is identifying clinical/business processes and getting data right.

Tony Shannon, Chair of the openEHR Clinical Review Board, triggered the discussion with this blog post - Sweden's noble eHealth strategy. Further emails elucidated links to all of the Swedish presentations from the Lund meeting, held on November 2, 2010, which I have included below.

However it was Sam Heard's email in that thread which summed up my concerns about the aforesaid 'elephant' and ensuring that health information is recorded consistently and safely. I particularly like the way he has described the problem, and his suggested solution. With his permission I post it here (with my emphasis)...

I was not able to get to Lund unfortunately - but it is clear there is some good work going on. In Inger Wejerfelt's introduction it was good to see the statement:

"Swedish national decision: Standards with focus on the information and not the communication only."

This is a critical shift and one that needs more attention. I cannot help but think we are stuck in a world where everyone thinks that health records can be whatever anyone dreams up and then they will interoperate. It is like the first few years of word processors and we are agreeing on what words are and paragraph breaks. We still have our own character representation and we are agreeing on sending paragraphs to each other to share. It is hard work.

Companies have made a lot of progress, Vista is open source and has a lot in it and GE's collaboration with Intermountain has the benefit of Stan Huff's insight. But health records are too important to do in a closed way or tied to a particular technology. Peter Fleming, head of NEHTA, has said "The Best of Breed is not an option" meaning we are not looking for good software that everyone can use. That is why I work in the openEHR community; so we can get on with this together and create a record that actually delivers what health care needs while allowing maximum diversity in applications.

A risk we might fall into is trying to do everything at once - this is not possible as 'everything' grows quickly from the previous everything. At this stage in developments the range of new possibilities transforms every year or less. So we need to decide on the first piece and implement it widely. For me that has always been the health record - concentrating on what has to be recorded. It will not do everything that is required but it will keep a record of it. Then we can then get smarter and do more.

The openEHR ACTION class embodies this divide between the model of care (workflow) and a model of recording. The care pathway steps are not actually the steps but rather the things that a clinician might record in following an instruction (or just doing it). People I am working with want to consider the workflow and transitions but the fact is that these all happen in the real world and do not get recorded consistently. When something _is_ recorded it is essential that we capture the state of the (possibly virtual) order. This means we have an idea of what is outstanding, ongoing etc.

We need more Swedens, more countries willing to think about the health record as possibly the most valuable resource. We need more people working on the shared logical specifications based on shared implementations.


Cheers, Sam

In Europe CEN 13606, now ISO 13606, has been mandated as the standard for EHR extract. A significant part of this standard is about the standardised definition of clinical content using archetypes. openEHR can be thought of as a superset of the 13606 standard, focused on the whole EHR, not just the extract.

    In particular:

  • the —UK, Singapore, Danish and Australian national eHealth programs are gaining momentum in determining EHR clinical content definitions.
  • the international openEHR community is gaining traction - clinicians defining clinical content through archetypes, and using the openEHR Clinical Knowledge Manager as a library and for life cycle management/peer review of archetypes, templates and terminology reference sets. Archetypes have been translated from English to Japanese, Chinese, Russian, Farsi, Dutch, German, Portuguese & Slovenian. They are also being created in other languages & translated into English. We are sharing clinical content models across international borders!;
  • applications based on shared archetypes are being built & implemented in Netherlands, UK, Australia, Brazil, Slovenia & the US;
  • professional clinical colleges are starting to drive archetype development in Australia & US to ensure that the EHRs they use are safe & 'fit for purpose'; and
  • Sweden's national approach is focused clearly on identification of the clinical business processes and defining clinical content through openEHR archetypes.

The Swedish approach is well documented and publicly available. The following presentations were given by key figures within the Swedish eHealth program at the international meeting held recently in Lund on November 3, 2010 (courtesy of Rikard Lövström):

  1. V-Tim - Inger Wejerfelt, Chief Health Informatics Officer, CeHis A presentation about V-TIM – the applied information model that is a framework for the “content and context” described for a clinical perspective and a result of national and regional projects.
  2. What are we trying to achieve? - Nils Schönström, MD PhD, Senior Advisor CeHis
  3. The Swedish eHealth approach - Dr Karl-Henrik Lundell
  4. The Swedish eHealth approach- Dr Håkan Nordgren, Head of National e-Health architecture board CeHis, MD .
  5. openEHR in Sweden and beyond - Thomas Beale.
  6. SNOMED CT - concept model attributes, archetype integration and terminology binding made by Jessica Rosenälv
  7. The reference archetypes - Helene Broberg
  8. Interdisciplinary Terminology for Health Care and Social Care in Sweden - Bengt Kron.
  9. Business driven archetype development - Jessica Rosenälv

The US approach gets lots of airplay because of the sheer volume of eHealth in the US in every dimension. However, in my humble view, it has significant limitations when it is so strongly focussed on the messaging and communication, at the expense of the health information itself. In a few years when the messaging paradigm is no longer sustainable, what then?

Addressing the issue of standardising EHR content will make exchange of health information simpler by orders of magnitude - drawing data directly from the EHR itself and being incorporated directly into the receiving EHR itself at the other end, ready to be computed upon without any interpretation. This is a foundation requirement for safe exchange of shared health information record, for data aggregation, for secondary use of data and for knowledge-based activities such as Clinical Decision Support.

Messaging hubs alone will not sustain the EHR revolution! Consider the other approaches that are also proving successful, especially that of the Scandinavians who are widely regarded as being the world leaders in this eHealth environment - they haven't got it perfect yet, but it is worth watching them very, very closely...