… I think the hype risks misunderstanding the problem that we’re solving here. Ultimately we’re not just sharing healthcare information between computers, but between people. We must be able to share data between people and organisations, and be confident that we know what others meant when they captured or authored that data.
I spent the week before last at the HL7 International Working Group Meeting in Baltimore. It’s an exciting time in the life of HL7, the international health standards organisation, as we witness the rapid maturing of the emerging Fast Health Interoperability Resources (FHIR) standard.
FHIR has been developed using modern web standards and architecture (i.e. JSON, XML, RESTful interfaces, etc.), with a strong focus on actually being understandable and implementable by healthcare software developers. FHIR also seeks to provide a model for the majority of concepts within the healthcare domain, allowing users of the standard to capture, store and share data with the same structure, and (to the extent possible) with a reasonably common view of its meaning.
The real genius of the approach behind the emerging FHIR standard is that it is being collaboratively developed by a growing number of real-world users. Previous healthcare data interoperability standards spent many years in an ivory tower development process, resulting in standards that were too complex for many people to use, with no significant opportunity for feedback from real world use before the standard was finalised. By contrast, FHIR is proceeding on the basis of real world use, with regular “connectathons”, ensuring that all implementers of the standard are able to exchange information with one another as expected.
And it’s working well. Really well. The market is very excited about FHIR and its potential where previous attempts at standards for healthcare data interoperability have largely failed. So excited in fact that you could argue we’re at the peak of the hype cycle. I hear some people talking about FHIR like its going to solve world hunger as well as health data interoperability!
So I wanted to hose that down a little with a dose of realism, as I think the hype risks misunderstanding the problem that we’re solving here. Ultimately we’re not just sharing healthcare information between computers, but between people. We must be able to share data between people and organisations, and be confident that we know what others meant when they captured or authored that data.
Sharing data between computers is easy. Sharing data between computers with a shared agreement on what that data means is hard, and requires high quality data and significant work to pre-agree what things mean. Sharing data between people is harder again. In fact, this is the main part of the problem, and is almost completely ignored by current healthcare data interoperability efforts.
You see, as we educate health providers through the medical education process, we are creating in the minds of our future health providers a conceptual model of healthcare. But from medical school to medical school, from country to country, there is no uniformity to this conceptual model. There is no architecture for healthcare that is educated into the minds of future doctors and nurses.
Let me give a simple example. What is the distinction between a “problem”, a “diagnosis”, an “allergy” and an “adverse reaction”. Health informatics specialists spend countless hours agonising about the distinctions between these concepts, whilst few real-world health providers have ever given this significant thought. Consequently, however well designed, sophisticated, pragmatic and complete a standard like FHIR becomes, its purity will always be violated by health providers who are not educated to understand the importance of data quality and use a consistent conceptual model of healthcare.
Until we address this problem as part of the medical education process, our efforts at health data interoperability will have mediocre outcomes at best. So I’m proposing a new course to be fitted into our medical education syllabus – “The architecture of healthcare”…
Now, who wants to engage me to come and teach this new course?