The openEHR Clinical Modelling Program commenced life as the Archetype Editorial Group and ‘operated’ between 2007 and 2012-ish. The original announcement in 2007 described it as the lead group for archetype authoring. Comprising a group of doctors influential in health IT and friendly to the openEHR approach selected by Sam Heard (Ocean, AU) in the first instance, the announcement also states the intent to develop a nursing group would be convened soon after.
Members that I can recall included Sam Heard, Sergio Carmona (Chile), Beatriz De Faria Leao (Brazil), Sundaresan Jagannathan (Jag) and Ian McNicoll (both from Scotland), Omer Hotomaroglu (Turkey, who went on to be the primary author for the ECG archetype) and Evelyn Hovenga (Australia, and despite being a nurse). Sam Heard gave me the title of ‘Convenor’, if I remember rightly, although it was never clear to me what the function of the group was, and I didn’t convene much. Effectively we had an oversight group without a clear purpose. I’d email them an archetype with some questions and, usually, no one responded. That wasn’t unreasonable in retrospect – if you haven’t any reason to be interested in the content of an archetype, then why would you spend time out of your busy day?
My main takeaway from this period was that expecting a committee of experts to provide editorial feedback about something that wasn’t relevant to their work or needs, especially without providing a framework for discussion was never going to work. We had to work smarter and be more focused on what we needed to achieve. This heavily informed the initial approach to clinician engagement and governance of CKM.
In 2007, an academic paper on the earliest thinking about clinical knowledge governance and describing the CKM precursor, the online ‘Archetype Finder’, co-authored by Sebastian Garde, Evelyn Hovenga and Sam Heard et al [1].
Sam Heard showed me some of his first archetypes in 2004 when I first joined Ocean - Adverse reaction and Blood pressure, if I remember rightly. But 2008 was the year that the first serious modelling work was done as part of an evaluation project for the NHS England’s infamous NPFIT program. Sam Heard and I worked in London for a number of months producing the first serious tranche of archetypes to support 2 use cases - the top 10 presentations to the Emergency department and archetypes to support antenatal care from a first visit through to a postnatal consultation. It was during this period that we realised that we needed the CLUSTER class of archetypes - they didn’t exist in the original specs. I still have those archetypes in an archive - interesting to see how our modelling thinking has evolved.
The first iteration of CKM development was developed during 2007/2008 - coded by Sebastian Garde, me as Product Manager, and with the first archetype uploaded around July 2008. For the next 10 years Sebastian and I battled awkward time zones to meet regularly, usually twice weekly via Skype, to plan, discuss, revise. Sebastian would code while I slept and vice versa. Ian McNicoll contributed regularly during the period when he was employed by Ocean.
Sam’s initial theory of developing groups with semi-autonomous, independent authoring and editorial responsibility for a content domain or speciality was attractive in theory, but the experience of the Editorial Group made it clear that this was not very practical. In response, the structure of the CKM was designed to support a folder-like structure – at the subdomain level as well as governed projects and ungoverned incubators to facilitate all kinds of groups with roles and responsibilities, and at various levels of granularity. This has worked quite well in practice - and is used extensively in the International CKM. Any work in projects is governed tightly by Clinical Knowledge Administrators, as these archetypes need to conform to the intent of a coherent ecosystem of archetypes, avoiding semantic overlaps and minimising gaps between concepts, always aspiring towards the ideal design each archetyped clinical concept being inclusive of any/all relevant data points and for any/all relevant use cases.
I have previously written about the State of the CKM in 2010.
From left Michael Beale (Ocean, UK), Rong Chen (Cambio, Sweden), Dipak Kalra and David Ingram (UCL, London), Ian McNicoll, Adriana Danilakova (Ocean, UK), Shinji Kobayashi (Kyoto University, Japan), Joana Feijó (Critical, Portugal), Seref Arikan (Ocean, UK), Jussara Macedo Rötzsch (Brazil), Sam Heard and myself (Ocean, AU). Photo credit: Shinji Kobayashi.
During an extended Board meeting held in London in September 2012, Martin Severs (not in the photo), the man who had recently coordinated the transformation of SNOMED into its new organisation, joined the meeting as a guest and he identified the Clinical Modelling Program as the single most critical success factor for openEHR. Everyone agreed that the programme should receive priority for funding from the Board.
At that point, at Martin’s request, I developed a proposed cost for a paid CKA at one day per week, just enough to kickstart the transformation of the CKM and modelling program from a special interest group towards a more formal and strategic approach. But no funding was able to be made available and unfortunately, this was the first and last meeting held about the strategy for the modelling program, at least to my knowledge.
From day one the Program has evolved, somewhat organically and based hugely on goodwill and a shared vision. We made do, growing and promoting the modelling approach the best we could with almost zero resources available from the fledgling Foundation. Ocean developed the CKM and provided it to the openEHR community for free, as well as covering my time and Sebastian’s. Others volunteered their time and effort, or their time was covered by their employers.
In building up and growing the Clinical Modelling program in the early days, I couldn’t find any examples of similar work to leverage - not on clinical modelling with the intent of building a coherent domain for the whole of health, nor associated clinical knowledge governance. We seemed to be in uncharted territory, armed with little more than the earliest Archetype Editor tool, an unstable flaky Template Designer, a fledgling CKM and a lot of enthusiasm.
If anyone has any additions or corrections, I’m happy to include or amend…
Next… the state of the Clinical Modelling Program
[1] Garde, Sebastian & Hovenga, Evelyn & Gränz, Jana & Foozonkhah, Shahla & Heard, Sam. (2007). Managing Archetypes for Sustainable and Semantically Interoperable Electronic Health Records. eJHI - electronic Journal of Health Informatics. 2. e3.