It seems this e-health business is much harder than we all thought

---

Recently, Dr David More was fortunate to spend time with a team of academics from Australia and the UK to explore some of the issues around the implementation of centralised National E-Health Systems, with a special focus on the UK, Canada, New Zealand, the US and Australia. Dr More writes that despite the difference in starting perspectives, the conclusions reached and the perceptions of what is needed seem to be remarkably aligned.

---

It was pretty clear from our conversation, and my own research, that there were very few, if any, top down public sector national systems that have not either failed or seemed to be heading that way.
Interestingly, the reasons seemed, in most cases, to be quite similar and to revolve around issues of the lack of any compelling value being offered by the proposed system to those who were the intended users; a considerable rigidity in approach from Government which was very much ‘one size fits all’; often politically driven and unrealistically short timelines; lack of genuine clinician involvement; a failure to distinguish between an IT project and a clinical project, as well as investment decisions being made on the basis of so-called ‘business cases’ which grossly exaggerated the potential benefits and typically substantially underestimated the costs that would finally be incurred.
In the broadest of terms there seemed to be agreement that if benefits were to be obtained for a National Health System through the deployment of Health IT the strategy most likely to be successful would involve the following:

1. 
   Focus on the absolute basics - key GP and specialist system capabilities along with secure communications between GPs, specialists, pharmacies,hospitals, allied health and service providers (labs, imaging, etc.).
   


2. Sticking to proven and well understood technical Standards. Health care at an operational level should not be operating at the ‘bleeding edge’.


3. 
   Initially working at a scale that is manageable and where delivery can be assured. I feel that the size of the old Divisions of GP or Medicare Locals (as they now are) seems about right to develop and plan a project roll out. (as they now are) seems about right to develop and plan a project roll out.
   

If these information flows can initially be established at a very simple document level and are working well there will be clear benefits to all parties in the information exchange and, assuming national Standards are used appropriately progressive scale-up of a solid, working Health Information Exchange would easily be possible.
An obvious step here would be to adopt the Clinical Care Record (CCR) Standard or similar to provide an information header that would assist with care co-ordination as it constitutes a valuable summary of health status and problems.
Equally, as the systems were fully bedded down and operational, the information content of the messages can be improved and made richer to provide greater clinical and patient benefit.
With all that working, consideration can then be given to developing improved access for patients to their information should they want it and improved services for patients interacting with their clinicians. Progress in such a structured way offers the best chance for really positive and useful outcomes for both clinicians and patients.
To have all this work well the only support required at a national level is the provision of three key infrastructural elements. There are:

1.  A Health Identifier Service - so it is possible to safely join pieces of information from different sources for the same patient.


2.  An End Point Location Service - to allow messages to be directed to the correct recipients. (These presently exist at more local levels).


3.  A properly led, funded and governed Standards setting mechanism at a national level.

To me, having these basics properly implemented and used is where most of the ‘pay-dirt’ is for all involved. Interestingly, all this can be done using proven and established technology and systems.
To move forward from here we need to really start to think how we can address the second issue that was a major topic of conversation. This is the issue of the astonishing complexity of representing clinical information and clinical thought processes in computerised form. This issue is a real sleeper in much e-health research because as anyone tries to move forward to better define clinical knowledge they soon realise the limitations of both language and the available coding systems to express and capture clinical thought processes and to structure and represent such concepts reproducibly. The semantics of clinical information turns out to be very complex.
The fact that complex messaging standards such as the full version of HL7 Version 3.0 and Clinical Terminologies such as SNOMED-CT are still works-in-progress tens of years after work began is a testament to this truth that capturing semantics and clinical knowledge is actually very hard indeed!
If you want a taste of just how hard all this is, I suggest exploring this project web site from a few years back http://www.semantichealth.org/. Work Package 6.1 written by Professor Alan Rector of Manchester University is especially enlightening.
A plea, from the trenches, to all bureaucrats and Governments, is to accept the big picture internationally and recognise large centralised systems are especially a career limiting and wasteful plan, while a focus on doing the bottom-up basics (as has been done pretty well in New Zealand) is a much better path for reduced cost, patient benefit and safety, clinician satisfaction and career longevity.