SIIM: Experiment in web technologies points to future of health IT

Though the standards that govern health IT may not evolve as quickly as some would like, a session on “Web Technologies” at the recent SIIM meeting in Long Beach presented evidence that DICOM and HL-7 are finally incorporating the fleet and favored representational state transfer (REST) application programming interfaces (APIs) that have transformed web-based banking and retail and now promise to transform healthcare.

“We’ve heard from the two main standards bodies that we interact with, DICOM and HL7, about how they are documenting and standardizing specific REST APIs,” Don Dennison, consultant and SIIM board member, told an audience on the first day of the meeting.

He then introduced early adopter Matt Coolidge, who described how his team at the Cleveland Clinic built a REST-based API in a proof of concept that is now moving into its second iteration—by demand. Coolidge is the associate director of development for the five-year-old Clinical Solutions Center at the Cleveland Clinic.

REST—defined in 2000 by Roy Fielding as a desired web architecture—identifies existing problems, compares alternate solutions and ensures that protocol extensions do not get in the way of optimal web communications. In contrast, health IT is characterized in most settings by clunky, system-to-system communications, responsible for many of healthcare’s interoperability barriers.

The Clinical Solutions Center was founded five years ago on the premise of bringing service-oriented architecture and REST APIs into use at Cleveland Clinic. An API can ease the work of programming graphical user interfaces, according to Wikipedia.

“APIs and SOA concepts, that’s where the future of healthcare is,” Coolidge asserts. Initially, however, those concepts had very little traction at the Cleveland Clinic, and Coolidge struggled to get resources. Working with a small team of two EMR developers, two web developers from outside healthcare and two clinicians, Coolidge began work on a proof-of-concept API that could pull data out of the EMR and put data into the EMR.

In 2012, the team launched its first REST-based API using its own data model. “We weren’t aware of the FHIR application at the time,” Coolidge explains, in reference to the new web-based version of HL-7, which stands for fast healthcare interoperability resources (FHIR).

EMR lite

After resolving security issues—like making sure that a patient’s social security number didn’t show up in a URL—the team’s first REST-based API was released in 2012. In 2013, the team opened the API to other developers, internally at first but later to trusted vendors. “Today, we are over 50 million calls, just in our REST API,” Coolidge shares.

Coolidge’s web developers had no healthcare experience, but a great deal of experience programming financial and consumer applications for the web. Coolidge himself came from a company that wrote online food-ordering APIs. Web-based programming has the advantage of being lightweight, which is useful in an environment that is increasingly mobile and plagued by network and connectivity problems.

In the web world, large communities build and work on APIs, resulting in rich and diverse ecosystems of applications and functionality. While the Cleveland Clinic team is small, it is already making its mark on the enterprise with a number of applications.

• IRIS, a rounding version of the team’s IOS application for physicians. Physicians can pull up their patient lists on their phones in the elevator on the way to the ICU to review vitals and images. Patient lists are stratified based on a risk score.

“We are not trying to fully replicate the EMR, we are trying to distill it down to an experience that meets the minimum requirements,” Coolidge explains. “It’s fast and easy for the users. It’s actually caused a problem because, now, all of the doctors are asking for it. We did an iPhone rollout and there are 2,500 iPhones in the organization. We can’t just roll it out to the whole enterprise with a small team. I think we will do a full roll out in the next year.”

A mobile vitals-entry application for nurses. This is based on the same framework and APIs as the rounding application, featuring some of the same views, but with a focus on inputting new values that are filed to the EMR. The application also issues alerts if there is a significant percent change since the last entry.

“We catch some mistakes and optimize the workflow,” Coolidge explains. “One of the latest features that the nurses were really craving was the ability to reorder some of that information based on their own requests. So, it is all filed in the same place, but they customize the UI to their preferences. We’ve seen a reduction in errors, and a reduction in missing documentation. There have been a lot of improvements with this, and we are continuing to study the implications.”

The Wellness Widgit, a web-based native application for the iPad. While the Cleveland Clinic’s EMR had some patient data-entry functionality, it was not simple, devices were not up to date and information was filed as a note, not discreetly, requiring the physician to go in and read all notes.

“What happens when someone has a high depression score and they may be at high risk of suicide?” Coolidge asks. “If the physician doesn’t review that note, there is no way to know.”

The application calls into the EMR using a web-service questionnaire and leaves a note that says the patient has not answered a specified list of questions, dynamically produced through the EMR. So that they didn’t have to build a questionnaire infrastructure, the team leveraged the built-in EMR tools and exposed the API. “That opened the floodgates of what we were able to do,” Coolidge says. “As they answer questions, exercise is a good example, and they say, ‘No I don’t exercise,’ then we don’t ask for any more information. If they say they exercise, we ask what kind of exercise they do, and all of this is filed into the flow sheets, into the system, and at the end, a thank you message from the head of the wellness program appears. The physician can review this inside the EMR. A very elderly Amish woman was able to use this and answer all of the questions—and she had never seen an iPad before.”

e-hospital, a Chrome application to monitor risk-score stratified. The application offers a full EMR experience through the synchronized opening of various EMR modules. “This is really important to keep straight, because these people are being monitored, sometimes in a distance-health situation,” Coolidge explains. “It is opening the eyes of the organization even further to what we can do with this.”

The developer of e-hospital had been with the team just one year and had no healthcare experience.

Moving forward, Coolidge and team are working on a second version of their REST-based API, with increased focus on the following four pillars: security, governance (including how they are creating new services and allowing access to those services), standardization (of data models and the protocols as well as creating a single developer portal); and infrastructure, making it “rock solid” as Coolidge grows his user and developer constituencies.

In the meantime, he is looking into changing older data models to a more consolidated data model, and a prime candidate is HL-7’s FHIR specifications. “That is something we’ve been looking at, and it’s a good option for us,” he says.