Why vendors are keen on playing with FHIR
In June last year, the senior vice president of medical informatics with US EMR giant Cerner, David McCallie, gave a speech to the company's annual developers' conference in which he outlined what he called a new approach that will enable plug-in apps to run natively inside any compliant electronic health record.
That approach is called SMART on FHIR, and is an extension of work that Cerner has been involved with since 2010 with the Harvard Medical School and Boston Children's Hospital.
Called the Substitutable Medical Applications and Reusable Technology (SMART) platforms project, the aim is to make it easier for medical apps to integrate into EHR systems at the point of care, and for EHR vendors to implement a common application programming interface (API) that will allow this to happen.
In 2014, the ability to do this was given a boost when the Fast Health Interoperability Resources (FHIR) developed by Australian eHealth expert Grahame Grieve and supported by international standards development organisation HL7 came on the scene in a big way.
The SMART team describes FHIR as a practical, robust approach to exposing granular data, which is needed for developer-friendly APIs. SMART has now mapped its technology to FHIR to produce SMART on FHIR, which it says will allow developers to integrate a vast array of clinical data with ease.
How the platform works in action was demonstrated at the HIMSS Interoperability Showcase in Florida last year, when Cerner showed how it could connect its EHR with a legacy system used by Intermountain Healthcare as well as the US Veterans' Administration's Vista EHR.
Dr McCallie told the Cerner DevCon audience that FHIR was the “fastest-growing, most exciting thing that has ever happened in HL7”.
“The uptake within the HL7 community when it was first put forward has been astonishing,” he said.
One of the most important elements of FHIR is the 'profiles' that have been described, which Dr McCallie said would enable true plug-and-play capability within and between big EHRs.
“The profiles are what enable plug and play,” he said. “It's one thing to have an API that everyone agrees upon but if when you move the data back and forth you can't understand that data, you haven't really achieved a whole lot of good.
“The profile describes enough detail about the data that is moving that you have what I call semantics by contract. SMART on FHIR will define 'app' profiles for developers.”
The lingua franca of data exchange
Cerner is just the largest of a number of vendors that are now beginning to play with FHIR. Closer to home, the Australian and New Zealand vendor and standards communities are also taking it on board.
One of those is Orion Health, one of the first major vendors to move beyond investigating the potential of FHIR to actually start on the road to implementing it. Orion Health announced last month that it would open up its Rhapsody integration engine to include FHIR capabilities.
Orion Health's product strategist David Hay has written a white paper on FHIR which projects that the standard will be in a 'normative' state by 2016-17, with future releases becoming backwards compatible.
Dr Hay believes FHIR is set to become the common language used globally to exchange healthcare data between disparate systems. “It is likely that FHIR will become like the Internet: ubiquitous and largely invisible – it’s just there,” he wrote.
“No one questions the use of the Internet and the standards that support it, and it seems probable that that could be FHIR’s future also – as the “lingua franca” of healthcare data exchange.”
Orion Health country manager Andrew Howard said FHIR promises to revolutionise the way health information is exchanged, and that the easiest way to explain why both the vendor and standards communities were so excited about it was that it would make exchanging data between incompatible systems much simpler and easier.
“If you track back to the early versions of HL7 v2 and the later developments with HL7 CDA and CCD, these are very difficult to understand and require months and months of training to really get proficient in coding those interfaces, constructing the messages correctly so they are meaningful and for systems to be able to interoperate effectively,” Mr Howard said.
“FHIR offers a different approach with the support of the standards community and the vendor communities at large. The adoption in approach to FHIR that all the major companies have taken is that we've all put our support behind the development of the standards before they have been formalised. Those investments are exciting the vendor community, exciting the development community, because they will simplify the way those systems interoperate.
“In layman's terms, we'll be able to build systems faster and share information in the health sector faster than we have been able to do before.”
Mr Howard said it was the ability to easily create APIs that can be used for plug and play apps within larger EHR systems that will allow faster and easier interoperability. Orion Health plans to expose the FHIR APIs initially through Rhapsody, a capability that should be available some time this year.
While Orion Health is one of the earliest adopters, Mr Howard believes most of the other larger players will support it as well.
“That's the excitement in the sector,” he said. “Orion Health's approach to the market has always been around interoperability and being open so you can exchange information. We all know that the health sector operates best when there is care co-ordination or care disaggregation, so with that philosophy in mind I think what you are seeing internationally is a recognition that healthcare systems need to be joined up.
“The larger vendors and even the smaller ones are moving towards that direction. There is no one or two dominant players in the market – there are always going to be multiple systems and that is why FHIR is going to be so important.”
The quick adoption of FHIR does not necessarily mean that it will replace v2 or CDA, but will work alongside those standards in the short to medium term, he said.
“What we have seen with v2 and v3, things sit alongside each other. v2 hasn't gone away with the introduction of CDA and I don't think you'll see the old messages disappear overnight. There will be a transition period, but the newer systems will start to pick up FHIR because it's a more efficient way to implement.
“With the data structure, v2 and v3 went a long way in terms of defining the way health data would be represented. Those things won't be lost with the implementation of FHIR.”
Primary care applications
While FHIR has got some of the bigger vendors on board, it also promises to be suitable for smaller vendors, including those working in primary healthcare. Melbourne-based health informatics and standards expert Brett Esler became interested in the potential of FHIR after Mr Grieve first explained his idea to the board of HL7 International in 2012.
Mr Esler, well known in the sector for his work with Pen Computer Systems for many years, set up his own company, Oridashi, in late 2012, and has since built an independent middleware platform called Hiasobi that uses FHIR to allow companion tools and apps to integrate easily with commonly used general practice clinical programs like Medical Director and Best Practice.
While primary care software vendors do not necessarily yet see a business case for making their systems interoperable, Mr Esler can already offer read-only interfaces through Hiasobi that can extract data from those systems in FHIR formats.
“That means you can query for information out of the systems and get it presented to you as FHIR,” Mr Esler said. “The benefit of that is obviously those two systems now look similar in terms of their information, so you can now start to write different applications for two systems instead of just one.”
In advance of moves towards interoperability between primary care vendors, this is where Mr Esler sees the early potential for FHIR and why he thinks it is eminently suitable for smaller vendors.
“FHIR is something that can be readily done and it actually makes implementing APIs simpler,” he said. “Some of the vendors have APIs already, but I would say why define your own when there's something that's standardised and well defined?
“To me it’s a good solution to that problem. And with benefits like ease of use, I think that really helps with the smaller vendors, as well as in terms of the cost of getting these sort of interfaces together.”
Mr Esler said developers working on client applications for mobile devices are also extremely interested, as it will make that work much easier.
“It’s much more aligned with how those sorts of mobile applications work and really well suited for that sort of business. A lot of the people that have those little apps, that want to do things off the side of any sort of system, they are very interested in FHIR.”
From a standards development point of view, rarely has Mr Esler seen a standard or specification move so fast from concept to draft standard for use.
“The exciting for me is that it is a working standard at every point, so it’s not just a specification,” he said. “But it's more than that: it’s a community of people that have actually implemented the current draft and worked through problems and solved problems and improved the specification as well.”
He said one of the main reasons was that it was easily understandable. “I call it normal development language. It’s a normal way of developing tools. It’s understandable, and one of the things that is really powerful is that it’s a complete definition straight out of the box.
“It includes not just how to represent the data, but also how to access to the data and query the data and all the technical reference material that you need to actually make a working system. It's much more straightforward to pick up than just data specifications and understand how it works and use it.”
Mr Esler said primary care vendors are moving slowly towards interoperability and standards-based implementations, and he would encourage them to look at FHIR in terms of what it can do.
“You'd have broad support across all the vendors and be able to have a very useful set of tools there,” he said. “And it would support things like some of the MSIA and RACGP programs around things like patient transfer, where you can take a patient record and move it from one system to another. Something like FHIR would be ideal for that.”
Klaus Veil, an adjunct associate professor at the University of Western Sydney and former health CIO who has been active in eHealth interoperability standards development and implementation since 1996, said that while Cerner and Orion Health are some of the earliest adopters, it is expected that most of the other vendors will soon support FHIR as well.
"What many of the industry commentators on FHIR overlook is that the approach of this new standard is a real paradigm shift," Mr Veil said. "The main reason for the astounding success of FHIR is that it is based on well-established information and communications technology methods, rather than custom developments.
"The second success factor is that FHIR combines the best features of existing HL7 standards with the latest web technologies. It also recognises the learnings from the failure of HL7 version 3 achieving implementation success, which is largely attributed to its complexity. FHIR is refreshingly simple.
"No health IT gurus are required for FHIR and even third-year university students can successfully implement working FHIR eHealth systems in a few weeks.”
Mr Veil is currently undertaking research into the factors that determine the adoption success or failure of healthcare data interoperability standards.
"The genius in Grahame Grieve's initial approach to FHIR in 2011, when is was still called Resources for Health (RFH), was to use computing technology that we use every day such as web browsing (HTTP), web searching (URL queries), online transactions (REST), logging in with Facebook or Twitter (OAuth2), online identification (openID), etc.
“These are well-known and well-established ICT technologies rather than the extremely specialised implementation methods and tools that HL7 had created for its v3 and CDA standards. In effect, FHIR has taken healthcare interoperability mainstream."
Mr Veil said he did not think that FHIR will replace v2 or CDA interfaces that are working well.
“However, I see that FHIR is rapidly becoming the interfacing method of choice, possibly relegating v2 and CDA to legacy standards, as was recently suggested by HL7 International CEO Charles Jaffe. Essentially, HL7 v2 may well become the COBOL of healthcare interoperability."
Posted in Australian eHealth