The journey continues…
One of the moderately successful incremental innovations that has given us some reprieve has been the development of standardised messages or documents to exchange selected, critical data – the ‘River of Limited Exchange’. There is no doubt that these messages have made a significant contribution to sharing health data but it is also important to note here that the transformation to and from each message or document carries its inherent risks to data quality and integrity.
One of the biggest issue with this approach is that the standardisation of each message or document requires negotiation and collaboration between each stakeholder – in the standards world these negotiations can take anywhere between one to five years to identify requirements and reach agreement on the scope and specification for each message/document clinical content payload. Negotiations between local organisations or vendors can be less time consuming but the work and effort required is still substantial. If this needs to be negotiated for every type of clinical communication between every stakeholder then we quickly find ourselves in the ‘Bay of Never-ending Negotiation’ and floating out into the ‘Sea of Unsustainability’.
And what happens when we want to exchange more than ‘selected, critical data’? The old mantra about the ‘right data, right person, right time’ is great rhetoric but we can’t do that if we are constrained to use minimum data sets or fixed messages/documents.
We need to work smarter! At the very least we need to lift our gaze up from our current plodding path and avoid hitting the Semantic Cliffs without a plan.
There have been attempts to escape the Swamp and scale the Cliffs towards Interoperability. Significant effort over the past 2-3 decades has been channelled into the development of clinical terminologies and there is no doubt that terminologies have been a key enabler for some of our incremental innovation success. On the international stage SNOMED CT is probably the frontrunner. However SNOMED’s evolution has largely been focused on creating another type of silo, a semantic silo that has mistakenly grown and expanded to try to represent all clinical things for all clinical requirements, enthusiastically supported by many of our standards development organisations and national programs. The core of SNOMED is immensely valuable, no argument here. But it is the drive to make this terminology also act as an information model that has arguably resulted in creation of a resource that is massive and unwieldy, often fragmented, inconsistent and so complex that to use it in its entirety is beyond the ability of most mortal clinicians or vendors.
We also know from our extensive experience with silo systems that proprietary information models have not been enough. Even if we try to share information models across vendors this will not solve the issue – archetypes as information models are not enough by themselves.
So how to reach the State of Interoperability?
Albert Einstein is reported to have said, “We cannot solve our problems with the same thinking we used when we created them.” While there is much debate as to whether this quote is correctly attributed or not, the importance of the message remains. Can we expect to achieve interoperability of data within any reasonable time frame if we continue to develop health IT solutions with our same incremental innovation approach. We have been working at this for over 30 years now and it can be confidently said that progress has been glacially slow. Isn’t this parochially known as the definition of insanity – doing the same thing yet expecting a different outcome? How much longer are we prepared to persist?
We are edging towards the Quicksand of Despair inexorably, day after day. We will eventually reach the barrier of the Semantic Cliffs as an impenetrable vertical barrier.
The answer is deceptively simple – we need to change our approach. Scary, yes. At the same time it is exciting. We have sensibly and logically pushed incremental innovation to very near its limits and now it is time to take a step back and reconsider, to try something new, orthogonal, and somewhat radical.
We need a Bridge of Smart Data! A shared and semantically accurate, safe and unambiguous data foundation that we use as the basis for all eHealth activities: for persistence; exchange; aggregation and analysis; decision support AND research.
The good thing is that we are not the first. This approach has been in development for over 20 years now, tested in international implementations and refined. This smart data approach has been pioneered within the openEHR Foundation, and is now in use within the international openEHR and a grassroots UK community, and by national programs in Australia, Norway, Brazil, Slovenia and NHS England.
But what about FHIR, I hear you ask? It packages information models plus terminology together as well, but is limited within the paradigm of a message standard. The problem of how to persist health data remains unsolved and the choice to limit standardisation of each FHIR resource to 80% of what legacy systems currently collect means that the furthest up the Semantic Cliffs that a FHIR implementation can reach is 80%, to the Plateau of FHIR.
By zeroing in on smart data we automatically change the focus of discussion from EHR applications, messages and documents back to its component building blocks – the re-usable granular health data itself – and once established the shared data specifications become the basis for the new generation of applications, messages and documents as well as all other health data activity.
By the term ‘smart data’ I’m referring to shared clinical content specifications which are:
- a combination of standards-based information models (such as openEHR archetypes) PLUS terminology knowledge resources (ranging from the high quality core of SNOMED CT down to value sets used only in a specific, local context)
- clinician driven:
- ensuring that the data we use represents how clinicians need their data
- unambiguous, safe and fit for clinical use
- collaboratively developed and agreed by domain experts, including but not limited to
- software engineers
- across an organisation;
- across a jurisdiction;
- nationally; or
- zero cost
- free access
- agnostic to any single technology or clinical system
- agnostic to any single terminology
- capable of multiple language translations
To scale the Semantic Cliffs we need to work collaboratively to build our Bridge of Smart Data. We have some components evolving – the collection of archetypes in the international CKM is growing in number as well as quality. Terminology is not so simple – the core of SNOMED CT will be a great candidate but is currently only available under license and this will limit availability to low income countries.
Most importantly, we need to confront the status quo NOW, and to start conversations that will help us to avoid the Quicksand of Despair and reach the State of Interoperability.
This is no longer just about a technical solution – we have implementations and over 20 years of experience in learning how to make it work. Now, we need to focus on the people, and especially the vendors, choosing to work collaboratively, actively deciding to break down the proprietary silos to focus on the data. Can we do it? This is our challenge.
Or rather, ask this question… can we afford not to do it?