It took nearly three years and three 510(k) applications, but MIM Software finally won Food & Drug Administration clearance for its Mobile MIM app, the first-ever mobile radiology app to receive the agency’s seal of approval.
The process was so fraught with confusion and miscommunication — or worse, no communication — that MIM CTO Mark Cain likened the experience to falling into a "black hole."
If there’s any good news for other radiology app developers, it’s that Beachwood, Ohio-based MIM’s experience might — might — provide a roadmap of sorts for the FDA to follow in reviewing similar apps for mobile devices like the iPhone and iPad in the future. And that’s no small point for a health IT industry that’s suffering from a lack of clarity when it comes to knowing what the FDA wants from developers of medical apps, many of which fall under the FDA’s jurisdiction because they’re considered medical devices.
"Clearance of this technology is a big step toward [the FDA] developing a model to review the next technology that comes down the pike. And there will be a next technology," said Bradley Merrill Thompson, a regulatory attorney with Epstein Becker & Green in Washington, D.C. (Thompson has written extensively on the intersection of regulation and mobile health technology, and his white paper (PDF) on the topic is highly recommended to anyone who wants to learn more on the subject.)
It’s not as if developing clear requirements for mHealth technology is an easy job for the FDA. The agency has already cleared a handful of apps, including one for heart monitoring and another for obstetrics, but a radiology app, whose success would be completely dependent on its ability to accurately present complex images, presented new challenges. Plus, the agency may be short on the experience and expertise needed to properly evaluate such a new form of technology.
An FDA spokeswoman declined an interview request because the agency is still refining its process for reviewing mobile apps.
"Clearance of Mobile MIM’s radiology app does not imply a generalized policy for mobile health devices," she said.
Still, the health IT industry no doubt hopes that clearance of the app will at least help the FDA formulate some sort of generalized policy for mobile apps. And with 80 percent of physicians expected to be using smart phones by 2012, the sooner that policy becomes clear, the better.
The Mobile MIM story
MIM Software has been developing radiology software since 2001, but as of early 2008, it hadn’t jumped into the mobile app market. That changed in March of that year, when — on a whim — two developers wrote the first lines of code for what would become the Mobile MIM app.
"It wasn’t part of our business plan. It just happened," said Cain. "Two of our employees began writing the code just to see if they could do it."
About two months later, the company had a working app, which shrinks the size of radiology images like CT and PET scans, transfers them securely and preserves image intensity values.
Knowing the app would be considered a medical device because doctors would use it to make diagnoses, MIM decided to make it available for free via Apple’s AppStore, hoping that the app wouldn’t draw FDA scrutiny because the company wasn’t trying to profit from it.
That’s not how things worked out. A few months later, Cain said, MIM received a notice about the app from the FDA saying, in effect, "Pull it now or our compliance department will be calling you shortly." The company wasted no time following those instructions.
"At that period of time, there were so many unknowns that the level of complexity was beyond our understanding," Cain said.
MIM then prepared and submitted a 510(k) application, which, if cleared, would have given MIM the right to begin marketing the app. From Cain’s perspective, the FDA’s response to 510(k) No. 1 left a lot to be desired. The agency questioned the company extensively about Apple’s iPhone hardware, about which the software company knew little.
"We got all these questions that didn’t make any sense," Cain said. "The communication breakdown was remarkable."
With both the FDA and MIM apparently unsure of how to proceed, the first bid for Mobile MIM clearance fizzled. MIM went "back to the drawing board," collected some clinical data and submitted it in another 510(k) in the summer of 2009. Radio silence. — MIM heard nothing from the agency, Can recalled.
"There was no communication, no feedback, no questions," he said. "We’d call and ask about it and we’d get, ‘No questions. It’s still under review.’"
That second application languished for 221 days. Then, in early 2010, the FDA told MIM that the bid would be rejected,a dding that the 510(k) route was the wrong protocal anyway. Lacking a "substantially equivalent" predicate device, the company would have to go through the more stringent pre-market approval program, which requires far more (and sometimes clinical) data and is thus a much more lengthy and expensive path.
After pondering its choices, MIM wound up appealing the FDA’s decision. In the summer of 2010, signs of progress began to appear. Reversing course, the federal watchdog agency said a 510(k) was the way to go after all, based this time on a prior version of MIM’s imaging software.
"There was a turn-around that summer," Cain said. "The communication was fantastic."
The company submitted 510(k) No. 3 in late 2010. The FDA cleared the app last week, stressing that it’s not intended to replace a full workstation but can be used to make a diagnosis when a physician does not have access to a workstation.
Cain expects the app to hit the AppStore this week for free at first. Radiologists will still to pay for the copmpany’s software to upload images to it.
Lessons learned
To hear Cain tell it, not much changed on MIM’s end. Throughout its dealings with the FDA, MIM made only one significant design change to the app, an adjustment to its "contrast response" that would compensate for possible hardware changes in subsequent versions of mobile devices, he said.
The majority of the FDA’s concerns all along were related to hardware, which Cain admits represents "a paradigm change" for the field of radiology. How does transferring a large image to a small screen affect a radiologist’s ability to discern subtle details? Is a consumer-oriented hardware device adequate for displaying complex medical images?
"You’re taking a consumer-grade instrument and you’re trying to show that it’s reliable enough to be used in a medical setting," said Thompson, the regulatory attorney. "There’s a lot to that."
Cain attributed the FDA’s ultimate decision to clear the app to two things: Personnel changes, which brought "new opinions and perspectives," and mounting pressure for the agency to provide more clarity to mHealth software developers.
"I’d like to say the [resulting] experience was purely because they decided to be transparent, but I think the pressure was on to finally accomplish this," he said. "It had gone on long enough. The medical world needed to understand the FDA’s viewpoint on portable technology in general."
The uncertainty surrounding the FDA’s regulation of mobile health apps is indicative of a "culture clash" between the Rockville, Md.-based agency and software developers’ Silicon Valley ethos, according to Thompson. The regulatory process is moving "very fast in Rockville terms, but very slow in Silicon Valley terms," he said.
"Silicon Valley is just not used to working in the public health environment, so it’s a difficult learning curve," Thompson explained.
At this point, there’s more pressure on the FDA than the software industry. The agency must fulfill its mission to protect public health without hampering the innovation that’s key for economic growth.
That means a host of questions without easy answers. What exactly, for example, should trigger FDA review of a mobile app? Should the FDA regulate apps that focus on general wellness, say for diabetes management or weight loss? If a networked device presents the wrong data, which vendor in that chain should be held responsible (imagine a mobile app on a device connected to a server, which in turn is connected to a doctor’s PC, which hosts desktop software)?
These questions will ultimately be decided by the agency, most likely after a period of public review. Until that happens, Cain said, there’s one practical thing developers looking to the medical app market can do: Make sure to have another source of income.
"For small developers, facing the black hole of the FDA process is tough," he said. "We were fortunate that we had an existing means to sustain our business during the drought."
