Dedication and Data Standards – the early days of FHIR

I love DevDays. It’s our annual chance to get together with the entire FHIR community and learn from each other. And this year I had the honor of interviewing the team who first put the FHIR standard on the map: Grahame Grieve, Lloyd McKenzie, and Firely’s own Ewout Kramer. We went on a fascinating trip down memory lane and reflected on the FHIR journey that followed. If one thing stands out for me, it’s the modesty and professional generosity of three people who have been at the forefront of some game-changing developments in healthcare IT.

Read on to be truly inspired and perhaps become a FHIR advocate yourself (if you weren’t already).

A radical reappraisal

First of all, I wanted to get a feeling of what it was like in the very early days, and whether any of them had guessed things would turn out the way they did.

It turns out that FHIR was a classic case of the right people at the right time. As late as 2011, HL7 was still focusing its energy on V3, which had clearly run its course. The organization started the Fresh Look project to brainstorm a future direction, and as a leading HL7 member, Grahame began his own radical reappraisal of what a future HL7 standard should look like. What he came up with – Resources for Health (RFH) – was the first sketch of what we now know as FHIR.

Grahame sent this strawman proposal for review to long term collaborators Lloyd McKenzie and Rene Spronk (now Firely’s product training manager). Lloyd was cautiously enthusiastic, but Rene immediately felt RFH had real potential to blow “fresh air” into the future of HL7 V3, and introduced Ewout Kramer to the others.

Ewout was working as a developer with V3, with increasing frustration, and immediately saw the power behind the concept. The three now began working on Grahame’s vision, with Ewout pushing for a strong developer focus. Ease of implementation was going to be crucial.   The embryonic team didn’t expect the idea to gain much traction at HL7. So they wrote a few webpages and waited for someone to stop them. When that didn’t happen Ewout and Grahame started writing reference implementations, and FHIR was on its way.

“The main barrier to overcome was ‘why should we do something other than V3’? And to be honest, that wasn’t a super-hard case to make to most people.” – Lloyd McKenzie

The baptism of FHIR

“RFH” was presented to the HL7 Working Group Meeting (WGM) meeting in San Diego in the fall of 2011. The HL7 Methods and Methodology (MnM) workgroup accepted Grahame’s proposal as a project, making September 2011 the formal birthdate of the new standard. After playing around with acronyms, the team settled on a name: Fast Healthcare Interoperability Resources (FHIR).

Breakfast in San Antonio

All three agree this was a seminal moment: the first planning meeting when they laid out the initial scope of the project, sketched out the resources and connections FHIR should include, and the work groups they needed. A Connectathon followed, and a third meeting, with fewer than 20 people, which laid the foundations of FHIR as Grahame, Lloyd, and Ewout began to create the FHIR manifesto.

“Our guiding principle was this: if software developers couldn’t make it work in a weekend, we knew we’d made it too complicated.” – Ewout Kramer

Since HL7 V3 was notoriously complex to implement, the manifesto put the focus firmly on ease of implementation. The new standard needed to:

• Focus on implementers
• Target support for common scenarios
• Leverage cross-industry web technologies
• Require human readability for interoperability
• Make content freely available
• Support multiple paradigms and architectures
• Demonstrate best practice governance

Laying the foundations

Creating a new standard is a huge undertaking. So how did they get from a blank sheet of paper to the first draft of FHIR? Can they still recognize the early work and identify their individual contributions in the latest versions? Ewout is quick to point out that the work groups changed things fundamentally, but some of his ‘pets’ are still discernible. The work he and Grahame did on structure definitions became one of the foundational pieces, and the data types Grahame created still endure.

“I knew from the beginning that profiles and value sets wouldn’t make sense in isolation – we needed to package them. And in the end, we got that.” – Ewout Kramer

From a band of three to a whole community

The San Antonio discussions proved that creating a proactive global community would be what really made FHIR different: so the team went ahead and did just that, building an open community and an open specification with its own self-sustaining magic.

Other development groups often ask how they can get ‘what FHIR has’, but its priceless spirit and enthusiasm can’t be retrofitted: it can only grow organically. And it felt like the community was just waiting to be created, building on a highly engaged and cooperative group of people at HL7 who were looking for change. Connectathons brought real developers together with experienced standardization practitioners, and by good fortune many of them were interested in both sides of the project and keen to get involved.  

By January 2012 the core team was firmly in place and FHIR began to gain significant traction in the HL7 community. HL7 reorientated towards implementers by expanding its tooling strategy to include ‘user tools’. The Firely .NET SDK reference implementation launched in June 2012 and FHIR Connectathons began the following January. The first FHIR DevDays took place in November 2014.

Before very long there were so many people working on FHIR that the number of concepts, and sheer engagement and enthusiasm gave it a life of its own.

“The success of FHIR is a team effort, but Firely has been there from the beginning, with outstanding development expertise and support for the community. They provide an open source library, and they organize FHIR DevDays. I think of them as ‘first among equals’. Firely makes an exceptional contribution to health equality.” – Grahame Grieve

Ten times better

Grahame, Ewout and Lloyd all had ‘day jobs’. They admit to working incredibly hard on FHIR in the early days – and continue to do so.

On top of their dedication, another key factor was limiting complications. What made FHIR a good idea was the notion of a single schema that’s the same throughout the whole world, focuses on the essentials, and is built on the same technology that scales the web.

To replace HL7 V3 it wasn’t enough to be better: they had to be ten times better. It took five years to get FHIR off the ground, with an expected life cycle of 15 years. And it’s now clear that a future replacement for FHIR would need to be ten times better in its turn – with a community that’s ten times as big. It looks like FHIR is here to stay.

“Does FHIR have a recipe for success? The developer focus and the community are crucial, but the main ingredient is that FHIR is simply a good idea. Especially the way you can build things on top of it.” – Lloyd McKenzie

Built on strong beliefs

Did the three co-initiators ever dream that FHIR would reach this point – mandated by regulators and with hundreds of people signing up for FHIR Dev Days each year? No, the dream was simply to build something better that would be useful in multiple jurisdictions, and they didn’t anticipate how much would get built on top of FHIR itself, such as SMART on FHIR and CDS Hooks. FHIR continues to grow in leaps and bounds and the community keeps drawing in new talent and generating new concepts.  

Grahame, Lloyd, and Ewout are proud of the amazing thing they made and the awesome things it’s achieved. Like an especially talented child, FHIR has a life of its own. All three are proud to have ‘birthed’ an open standard that’s making the world better for everyone, and not just in the first world. In the words of Grahame Grieve’s teenage daughter, “Dad, you’re a hero”.

“I’m proud that we built an open standard that’s making people’s lives better. Not just in the first world, and not just for the rich. A meaningful concept, that’s what it is to me.” – Grahame Grieve

Where next?

As the years go by the FHIR community becomes even more welcoming to participants from all over the world. The FHIR standard is stronger than ever, due in part to ONC and CMS regulations in the US, and the introduction of the ISiK rule in Germany and the MedMij standard in the Netherlands.

Ewout Kramer still plays a key role in enlarging the community and contributing to the standard, along with his Firely colleagues. Grahame Grieve is HL7 Product Director for FHIR, among many other roles, and Lloyd McKenzie is a consultant to the ONC and FDA, and HL7 ‘Facilitator at Large’.

I could have listened to them for hours, but I had one last question. With what they know now, would they do it all again?

I think you already know the answer.