In August, the Indianapolis-based Regenstrief Center for Biomedical Informatics announced that it was piloting the use of the Fast Healthcare Interoperability Resources (FHIR) standard to merge data from individual electronic health records (EHRs) with those stored in the Indiana Network for Patient Care (INPC), Indiana's common framework for health information exchange (HIE).
In the press release at the time from Regenstrief, a medical and public health informatics research organization that has been at the core of health IT innovation in the Indianapolis region for 40 years, officials brought up the following question: “You are rushed to a hospital in an emergency, is your complete medical record available to those caring for you? Will they know all medications you have been prescribed and whether you are taking them as directed? Does your primary care physician know your complete medical history?” Indeed, according to clinician-informaticians of the Regenstrief Institute, the answer to these questions is almost always "no." Not having complete health information available often results in subpar care and can endanger patients.
To this end, healthcare technology innovators and developers have pointed to SMART (Substitutable Medical Applications & Reusable Technologies) on FHIR implementing an open architecture to support interchangeable developer-friendly APIs (application program interfaces) that can be “plugged in” to any compliant EHR or health data container. Part of the allure is that the effort can get away from document-centric approaches and expose discrete data elements as service.
Enter this FHIR pilot, and with it, Titus Schleyer, M.D., Ph.D., a Regenstrief Institute investigator and Clem McDonald Professor of Biomedical Informatics at Indiana University School of Medicine, who is at the center of this project. "Imagine that you as a patient can use an ‘app’ on your smart phone to reconcile the multiple lists of medications maintained by several care providers into one authoritative, current list. And then, you can bring that list to your colonoscopy screening appointment for review by your physician prior to the procedure. That is huge, which is why the federal government is also focusing attention on helping patients do that,” Schleyer says, speaking to FHIR’s benefits.
Schleyer says it was about three years ago when Graham Grieve, the co-creator of the FHIR standard, approached Regenstrief about its potential. What happened next? What will the pilot look to accomplish? And what’s in store for FHIR’s future as a healthcare standard? Schleyer discusses this and more in the below interview with Healthcare Informatics.
What are the main details behind this project’s conception?
FHIR is not really an established standard; it’s kind of young. About three years ago, Graham Grieve said that it would have a very different dynamic than most other healthcare standards. He asked if Regenstrief would support it. We signed off on it, and while I have been in healthcare standards for 25 years, there’s a different flavor to FHIR when compared to other stuff. It was a risky decision for me, but in the past three years I have not seen anything in the healthcare standards [space] that has moved as fast and has moved with such great dynamic as FHIR has. It has “legs.” So how can we solve some of the longstanding interoperability and information aggregation problems using this new standard? That’s where the pilot came in.
Why was it a risky decision for you?
You never want to be the person who bets on the next Betamax. I was alive when that went down, and you heard stories about how VHS won against Betamax despite not being a superior technology. So you have to look at these new developments and wonder if you are betting on what will be there in five years. You do have to educate yourself about the technology and the approach, and you build stuff on top of it. During my life I have done a few things where I bet on computer technologies that I thought were great and a few years later the company that made the tool was out of business. But as I talked to Graham and other early developers, it was evident that FHIR had a quality user-centered design, which means you design something with the user in mind, getting in the head of how they operate and think.
What else about FHIR sold you?
With versions of HL7 Reference Information Model (RIM), you needed this huge level of expertise and you had implementation guides that ran thousands of pages, but FHIR didn’t have that look and feel to it. When we did this pilot in April, there was a “Connectathon,” in which 17 teams from the Indianapolis area, many of them having their first exposure to FHIR, were gathered together to [run demos] on it. I was in the room with the developers for two days, and by the end of the weekend these people were doing stuff that was really cool. With HL7 RIM, you’d have to lock these same people in a room for six months before they would have produced something on that model. It was amazing to see these 17 teams all turn out something that worked by the end of the weekend. So I thought, if the uninitiated can work with FHIR so quickly, it will be a winner.
The Connectathon was Epic-focused, meaning the event was about Epic with a sandbox and FHIR on top of it. This was one option for innovation; the other came about in the months prior when one of my faculties asked if we could offer FHIR on the INPC network. I turned to him and asked, “We’re that far along?” He said yes, so we now had two options in the event: working on Epic exclusively or taking the Epic sandbox and connecting the information from there to the same patients in the INPC. You can really do interesting things with merging data from two different sources. So in the end, seven of the 17 teams decided to showcase solutions that drew data from both sources. And of course we made sure the patients were the same; we could connect them using a unique patient identifier. I rarely say this, but I haven’t seen another place in the U.S. that used FHIR to integrate information from an HIE to a standard EHR.
Is there too much optimism with FHIR? Should people be tempering expectations some?
I’ve heard those same viewpoints, and I do tend to be skeptical of the cure-alls for everything. They tend not to materialize. But, hype does have its own dynamic. People who don’t know much about say it will solve everything. Even Graham says it won’t solve everything, but he does tell you what it will solve. We have an interoperability solution that takes the grunt work and heavy lifting that we used to do on the back end, and reduces that to a minimum. When you connect various computer systems to one another, you have to route HL7 streams between those systems and make sure that the receiving system can interpret what’s coming across that wire. And if the interface breaks because of an upgrade, you have to fix that again, and that’s the ugly part of interoperability. Yes, you can make it work but it takes time and maintenance to keep it working. With FHIR, you have a much more generic interface, as it uses web technologies that pretty much every web programmer is currently familiar with. It has a simple and powerful model of getting at the data.
With FHIR, you have FHIR resources and profiles, and they give you data in a predictable and understandable way. If you look at what FHIR puts out, someone who knows nothing about programming can read what’s coming across the wire. One of the dangers I do see is that vendors have a large leeway of what profiles and resources they implement, so if your expectation is that you will get the same functional or data access scope from vendor across vendor, you will be disappointed since not every vendor implements every single thing, as it depends on customers’ needs. That slants implementation some. But I talk to vendors a lot, and I see vendors aware of this danger and actively working to not let it happen. Here, the Argonaut Project is a collaboration around FHIR profiles. So you have the major vendors who are normally out to kill each other working together to make it consistent, so it’s predictable and useable for those who want to work with FHIR.
What do you make of the work that has been done with FHIR so far?
If you look around and at the SMART on FHIR collaborative, the energy now is mostly around building innovative apps on top of existing EHRs. These apps typically use the same raw data the EHR uses itself, but they do more interesting stuff. For example, there might be a pediatric growth chart that has made the rounds, and it’s an app to figure out if this child is progressing normally compared to what he or she should be at in terms of growth. So FHIR gives users and health systems the ability to take the data inside the EHR and do new things with it. There is not too much around interoperability yet, but more about adding value-added functionality to the EHR.
When I talk to the EHR vendors, all of them will admit—some more readily than others—that they cannot build all of the innovation that healthcare customers need. And this is the big driver for these open platforms, which we have been seeing in healthcare. Vendors are open to giving others the ability to innovate, and the more innovation that exists on their system, the more their platform becomes sticky. Microsoft Windows has been very focused on making sure everyone can work on its platform and develop applications on the platform, because that ultimately drives operating system sales. EHRs are now thinking of themselves as operating systems, and the more innovation you have on the operating system, the more attractive you’ll be to people.
What’s next for Regenstrief and FHIR?
Similar to the vendors, Regenstrief has unfortunately been a bottleneck. We have 15 faculty and 50 staff, and there’s only so much innovation you can do in an age and context when innovation is highly in demand. For me, FHIR is an enabler of an innovation community rather than an innovation institution like us. So what I’d like to see is FHIR becoming an enabler of innovation for health systems, for institutions like us, and for universities, where we increase the rate of beneficial innovation in healthcare. I talk to local health system CEOs regularly; we are hoping we can convince people in decision-making roles that this is a winner and you won’t have to wait five years to get 10 percent of what you really wanted. But, at the same time, let’s not oversell this and let’s temper expectations.