Looking Beyond Randomized Trials: Could a Wider Range of Studies Support Device Approval?

Two US Food and Drug Administration (FDA) employees are highlighting the different paths to approval for devices and drugs, arguing that in some cases, medical innovations should be held to a different standard of evidence, opening the door to a wider variety of supporting studies.

At least one naysayer, however, is sounding a note of caution, saying such a move would be good for industry but not necessarily for patients.

Although evidence from gold-standard randomized controlled trials is called for in many cases, data from other sources may be sufficient to establish the safety and effectiveness of devices, Owen Faris, PhD, and Jeffrey Shuren, MD, JD (FDA’s Center for Devices and Radiological Health [CDRH], Silver Spring, MD), write in the April 6, 2017, issue of the New England Journal of Medicine.

The aims of the article, Faris told TCTMD, “were really focused on education and outreach” as means of increasing familiarity with what device trials do and do not offer. “The clinical community tends to be more familiar with drug trials than they are with device trials, and we wanted to highlight some of those differences,” he said.

“Device trials are unique in many ways and really require us to be very flexible in the kinds of data we see,” Faris added.

When large-scale trials are not possible or feasible, the agency might turn to preclinical evidence from animal studies or bench testing, computer modeling, and real-world clinical registries, including those from outside the United States, the co-authors explain. Faris serves as the CDRH’s clinical trials director, and Shuren as its director.

“Those sorts of tools really offer the potential for clinical research to be faster, less expensive, and importantly just as informative—and in some cases, even more informative,” Faris said.

He said a new effort called the National Evaluation System for Health Technology (NEST), which is still getting off the ground, is designed to make better use of the clinical data being collected everyday around the country. It will link information from clinical registries, electronic health records, and medical billing claims that can be used to inform regulatory decisions. “It’s really trying to coordinate a learning healthcare system,” Faris said. “The idea being that a lot of information is gathered over the course of normal clinical care that can be informative to regulatory decisions and right now we use that suboptimally.”

That initiative has potential not only for postmarketing surveillance, which is crucial for detecting potential safety issues that could not be identified before approval, but also for premarket decisions in cases where an approved device is being used off-label and may be in line for an expanded indication, he noted.

As tools like NEST and computer modeling “develop further we hope and expect that our ability to answer essential questions about safety and performance and effectiveness will be more efficient and that important technologies can get to patients more quickly,” Faris said.

A Cautionary Note

Asked to comment on the article, Rita Redberg, MD (University of California, San Francisco), a long-time advocate for patient safety and better safeguards against aggressively expedited device approvals, was blunt. “If I was a venture capitalist who invested in the medical device industry or if I worked with the medical device industry, I’d be very happy. But I’m a cardiologist who takes care of patients,” she said. “I have to make decisions on whether to recommend medical devices, and I have to know whether they’re safe and effective for those patients. . . . Not doing clinical trials, which is a lot of what’s proposed in there, is going to make that job much harder to assure safety and effectiveness for patients.”

She cautioned against being too quick to move away from the idea of performing randomized controlled trials and depending on computer-simulated data to evaluate high-risk devices, pointing to what happened with the Wingspan stent, which was initially approved on the basis of a small study and later shown to be associated with worse outcomes compared with medical therapy alone in the SAMMPRIS trial.

“If you’re talking about a high-risk medical device that’s going to be implanted in a patient, I want to know how it works in patients,” Redberg said. “I don’t think a computer simulation’s going to answer that question. And as a patient, I would be concerned if I was essentially the test subject and it hadn’t been tried first in clinical trials.”

Redberg supported, however, the authors’ discussion of more robust postmarketing evidence collection.

“Certainly even if you do clinical trials, they’re still more limited in duration and smaller in sample size than we can ever get in postmarket data,” she said. “So clinical trials are important for establishing that a device is better than the alternative, but they’re not going to give us safety signals when there are small numbers of events.”

Also, she added, any registries set up for that purpose must be publicly available, providing access to information on safety events, as well as designed to collect information on all patients, not just a select group. “That’s not happening now,” Redberg said.