A study published in JAMIA Open, led by Regenstrief Institute Research Scientist Titus Schleyer, D.M.D., Ph.D., surveys the landscape of FHIR apps, providing a snapshot of how the standard is being used in clinical settings and for research. In an interview with Healthcare Innovation, Schleyer offered some observations on the evolution of FHIR-based apps.
In addition to his role as a Regenstrief Institute research scientist and professor of medicine at Indiana University School of Medicine, Schleyer co-directs the Regenstrief Institute-IU Richard M. Fairbanks School of Public Health’s public and population health informatics fellowship program. He also is the program director for learning health informatics at Regenstrief and leads the Indiana Learning Health System Initiative.
Schleyer said he has been in the FHIR space for quite some time, and watched it grow over the years. “I thought that it's great that so much is happening, but what exactly is out there? I had the idea and recruited a bunch of people, and then we worked for a year and a half and identified 119 apps in our study. It's like a mosaic — you add another piece and over time the picture becomes clearer.”
The study identified 74 clinical apps and 45 focused on research. Schleyer noted that the overwhelming majority of apps target healthcare professionals, and of those, the overwhelming majority were clinical. As far as diseases targeted, cancer, diabetes, cardiovascular, and critical care are all big, he said.
On the research side, FHIR is being used for making clinical trials easier, among other things. He said one of the apps that won a FHIR application competition a couple of years ago was a connector to get data into REDCap from clinical databases. “That was developed by colleagues at Vanderbilt, and that is a really attractive product because you can tell the research coordinators they don't have to re-enter data that were already in the electronic health record,” he said.
Some of these apps studied were developed by academic medical centers and others by software vendors. “People in academic places were definitely a substantial cohort of developers, but not everything that gets developed in academia makes it long term as a product,” Schleyer said. “There were a lot of small companies that developed these apps. Interestingly, most apps were actually not from EHR vendors. They were from the developer community, which tells me that FHIR lives up to that expectation of enabling innovation in many, many other quarters. It compensates for the fact that EHR vendors can't innovate on everything. By increasing the ease with which we access data, it makes it easier for other people to come up with innovative apps. That's quite evident from our survey.”
One purported benefit of FHIR apps is that they should be somewhat plug and play and not require a lot of customization. I asked Schleyer if they were seeing apps being deployed across health systems. “This is a major sales point for FHIR, but I would append an asterisk,” he said. “If you go from Cerner to Epic to Allscripts to eClinicalWorks, they are slightly different in the way the standard has been implemented. Sometimes that's easy, depending on what the resources are that you use, and sometimes you have to do a little bit of rework. But the good thing is that redeploying FHIR apps in multiple places really does save a lot of effort, even if there's a little bit of customization involved.”
The researchers also found quite a few patient-facing apps. Apps for blood pressure capture, diabetes management, mental health treatment — those are kind of apps that make sense to expose to the patient, Schleyer said. “For instance, people are struggling with managing their diabetes. If you don't manage your diabetes, well, your quality of life isn't great. So it makes sense that in spaces where patient engagement demand is high that there will be apps that help the patient.”
The paper noted that other researchers had conducted a survey on implementation of the SMART HL7 bulk FHIR access API, which has a focus on population level health data. Schleyer said this is a relatively new area of FHIR development.
“I would say, definitely, it's an area where I would expect significant growth,” he said. “The first goal of the FHIR standard was to make information access easier on a patient-by-patient basis, but eventually we realized we could use this to do studies. Well, then it's not practical to ask for one patient after another. That's why the standards developers who drive FHIR came up with a way to do bulk download of multiple patients."
Schleyer said the intersection of FHIR and artificial intelligence may also prove important. “One of the things that you can use the bulk FHIR API for is to ingest data into machine learning algorithms at scale, and then you can train your algorithm,” he explained. “The implementation of AI and machine learning algorithms also would benefit from lots and lots of improvements that we can make to the FHIR standard, so I think it's kind of a beneficial relationship.”
Schleyer has had his own experience developing a FHIR app called Health Dart, which helps emergency department clinicians access valuable patient information from the state health information exchange within the EHR. It is running in all 15 emergency departments at IU Health, and there is interest from other health systems, he said.
“One of the key things with FHIR development is what resources are you going to use? And are your source data mapped sufficiently and correctly to those resources? I'm not talking about the patient resource or the appointment resource — those are all fairly cut and dry. But the observation resource, for instance, is a huge resource. If you want to interrogate lab tests, you are interested in what code systems they are using and so on. That's definitely a challenge. Another challenge is if you develop a FHIR app, you want to go into an app store such as Epic’s. It takes a lot of work to comply with all the rules and regulations. You have to cross a lot of T's and dot a lot of I’s. Also, the FHIR standard sometimes doesn't cover all the use cases that you want, or maybe it does and the vendor didn't quite implement it the way you need in a particular moment in time.”
As a follow-up, I asked Schleyer whether the EHR vendors in general had done a good job of working with the FHIR app developers so far.
“What I'm encouraged by and that was not obvious five years ago is the degree of acceptance and adoption of the FHIR standard by EHR vendors,” he said. “You really have a tough time nowadays to find an EHR platform that doesn't have FHIR in it. They definitely see the benefit of FHIR. A very generalizable lesson in innovation was learned by the EHR vendors. If you go back in computing, Windows XP was probably one of the first major platforms to showcase that Microsoft positioned Windows as a development platform. They babysat their developers. They made it easy. By encouraging the development of an ecosystem of lots of interesting and different apps, you gain so much sustainability as a platform. If you think about it, Android, iPhone, Linux, all the strong platforms really have a very rich ecosystem of innovation that they enable, and I think the EHR vendors have seen the same thing — that FHIR, in the end, is good for them. You know, it's like life insurance for them.”