As I leave my leadership role in the Clinical Modelling Program, let me start by sharing with you some facts and insights…
As of 2 November 2021, from CKM statistics:
Clinical modelling community
All insights into the clinical modelling community is represented by those registered in CKM.
2663 registered users from 104 countries.
1015 individuals have volunteered to participate in archetype reviews - that’s 38% of the total number of registrants. This is the pool of people who we can draw on to invite to an archetype review. It does not represent whether they have been invited, nor if they have responded to an invitation.
300 (11%) have volunteered skills in translation.
Active archetype statistics, projects only
It’s been difficult to get a consistent handle on these as CKM has some inconsistencies in calculations but the trends are sound.
All active local archetypes, projects only – ~519
Draft – ~365
In review – 30, counted (6% of 519)
Published – 139, counted (27% of 519)
Review suspended – 7
Reassess – 5
Data point statistics
· All active archetypes, projects only - 5915
· Published archetypes - 1478 (25%)
Language statistics
Number of languages - 31
Top 5 languages
Norwegian Bokmål – 239
German – 206
Portuguese (Brazil) – 151
Swedish – 112
Arabic (Syria) - 78
Roles
Clinical Knowledge Administrators – 2
Editors
Archetype content
Regular - 5
Guest Editors on a special interest or per project basis -~15 have had varying levels of training/mentoring
Translation – 20+
Compare this with ‘State of the CKM’ in 2019
2343 registered users from 96 countries
857 reviewers
281 translators
478 archetypes
92 published (11%)
And just for reminiscing about the good old days – ‘State of the CKM’ in 2010.
Archetype numbers -Analysis and discussion
The Clinical Modelling Report presented at the recent 2021 AGM represented more about the dynamic nature of activity in the 12 months to August that is not evident from the stats above:
73 newly created archetypes – mostly by the Editorial team or via CKM proposals
155 archetypes modified/corrected/refined
30 archetypes newly published or republished
31 archetypes have undergone 47 review rounds due to 77 unique reviewers contributing 468 reviews (an average of 6 reviews per reviewer)
The top reviewer completed 39 reviews.
I think most people will agree that this reflects a dynamic and active Editorial and reviewer community participating in the CKM hub.
BUT let’s do a quick reality check…
The first archetype was uploaded in mid-2008.
In 13 years, we have published 139 archetypes – that’s an average of only 10.7 archetypes per year.
In the past 2 years, we have had a net gain of 41 archetypes over the past 2 years – mainly acquired through proposals, the COVID collaboration and projects such as the Genomics work program. Many old patterns or concepts were retired. The CKM has undergone quite a big clean-up to ensure it reflects the evolving modelling patterns, always aiming to create a coherent ecosystem of models.
In that same timeframe, the number of archetypes published has increased by 47, and the corresponding percentage rose from 11% to ~27% - that’s an average of 23.7 archetypes per year. This sounds positive, but is it really? Is this publication rate enough? What numbers should we be aiming for?
In the past, I’ve estimated that if we design archetypes well, 50 core archetypes would support most of a primary care EHR, maybe 2000-3000 to support all the clinical requirements for a hospital EHR. These estimates are probably still relevant. Although as we start to explore standardisation of secondary data sets that can be abstracted from the original EHR records – for purposes such as research, registries, and reporting – numbers will blow out even further.
Let’s use 2500 as a working number for the required number of archetypes to support an EHR.
At historical average rates, it will take us 233.6 years to publish the number of archetypes that we need. At rates from the past 2 years, it will take us 105 years. That’s clearly not realistic or sustainable. And if we add in archetypes for standardising secondary use, reporting, research etc, this balloons further.
What about the 365 draft archetypes that we already have? At historical rates = 35 years; at recent rates, 15 years.
Let me be perfectly clear here, in my opinion openEHR has the potential for becoming the ‘universal health language’, a lingua franca for health IT.
In order to transform digital health we need a common information model, and the best candidate we have to achieve this is with openEHR… This is what I’ve been working towards for over 15 years - a common domain information model supported by a coherent set of archetypes as the foundational clinical models.
But the reality is severely lagging, held back by a Board that has been absent and uninterested.
____________________
At the AGM, the CIC Board has proposed ‘refreshing’ and ‘rethinking’ the Clinical Modelling Program, apparently by going to vendors to request more models…
We have more archetypes than we can cope with at the moment. While gathering more archetypes will help to fill the content gaps, the rate-limiting step is actually the review and publication process. And this has been ignored for more than a decade.
The Board doesn’t need to ‘rethink’ the Clinical Modelling Program. It does need to stop ignoring it and start supporting it.
At the simplest level, it needs:
to understand the work of the Clinical Modelling Program (It thinks it does, but it doesn’t)
to understand the skills required to manage a Clinical Modelling Program that builds archetypes to support a coherent data ecosystem (No, despite the rhetoric, it absolutely doesn’t)
to develop a joint strategy with the CKAs who know how to protect the quality, integrity and credibility of the international CKM as the ‘source of truth’ for openEHR archetypes;
to identify and source adequate funding to:
train and upskill Editors and Clinical Knowledge Administrators to ensure that the current asset is independent and sustainable
ensure that the CKM can continue to function as an independent entity that underpins all openEHR implementations
a succession plan
The Clinical Modelling Program has effectively operated as an orphan within the organisation. It is more by luck than strategy or design that we have achieved what we have. This needs to change or openEHR could fail.
This is the kind of thing I wanted to speak to the Board about each time I approached it this year. Each time my request was rejected, with the reason being that Silje Ljosland Bakke had been appointed the Clinical Modelling Program representative on the Board at some point. Silje (already on the Board as an organisational rep) has repeatedly denied that she fulfils this role. So what on earth is going on. Certainly, the appointment of a Board representative has never been discussed with me. Rather both Silje and I feel that the Modelling program could only be fairly represented if both co-leads were involved in any Board discussions or program strategy.
One has to wonder, is this a consensus position of the Board as a whole, or it is just the position of the Co-chair who responds to my emails, who hasn’t spoken to me once during my period of paid engagement, not to understand the Modelling Program, nor even to enquire why I withdrew my services.
It’s messy and ugly and totally unnecessary. And the openEHR community is the ultimate loser.
More to come…