House Health Subcommittee grills FDA leaders over med device missteps

Despite the highest agency user fees on record, and granted new authorities to reform and accelerate medical device and biologics approvals, representatives questioned delays in getting products to patients faster.
By Andrea Fox
10:58 AM

Photo: Sarah Silbiger/Getty Images

During the House Energy and Commerce Health Subcommittee hearing on the U.S. Food and Drug Administration's regulation of drugs, biologics and medical devices on Wednesday, representatives fired questions at FDA directors. 

The lawmakers said they were aiming to examine how FDA rules bar doctors and patients from accessing tests that would need to undergo lengthy premarket approvals and how the agency is approaching reviews of artificial intelligence and machine learning-enabled devices. They also asked about third-party certification and other topics. 

Several committee members also pressed the FDA directors to improve on communications with applicants.

Too slow for innovation

Health Subcommittee Chair Brett Guthrie, R-Ky., said during opening remarks at the Congressional hearing Wednesday that several missteps have caused "uncertainty among innovators" on the bipartisan supported pathway.

"The repeated attacks on the accelerated approval pathway from CMS, which is second-guessing FDA's equities to academics and payers alike, insisting that these approvals are unproven or lesser than traditional review pathways degrades public trust," he said.

"FDA cannot move backward," but in 2023, "all three centers failed to meet critical performance process and hiring goals despite all-time highs in funding," said Energy and Commerce Chair Cathy McMorris Rodgers, R-Washington.

Stakeholders also reported that FDA staff is "more disconnected and difficult to work with than ever before."

"Everyone wants the FDA to succeed," but the committee wants to know what challenges FDA is facing and why, and how Congress can help, she added. 

To attract more innovation, Dr. Jeff Shuren, director of the Center for Devices and Radiological Health, said FDA launched its Breakthrough Devices Program that is now designating 100 innovative medical devices per year. 

The program seeks to identify and provide patients and healthcare professionals with access to devices that can improve treatment or diagnoses. The final guidance, released in September, states that the FDA may consider improved access to a device while determining if it meets the first breakthrough standard. 

However, most of the devices aren't getting to patients, because there are a number of obstacles going from concept to commercialization. 

"It is appropriately called the valley of death," he said.

A pilot program where FDA advisors engage proactively and strategically to work with innovators and identify their challenges, such as payer reimbursement, and Shuren said the feedback has been positive. 

One company said they saved a year in development by participating, he told the committee.

Need for third-party AI-certification

The FDA has also approved 900 AI/ML medical devices, according to Shuren. 

However, a third-party certification model needs to be in place to advance innovation, and for that the agency needs help from Congress, he said. 

Shuren, CDRH director for 15 years, said that, initially, novel technologies were not being offered to the United States, and the agency made changes to attract innovation. 

But by 2023, there was a "five-fold increase" in novel technologies, with 124 approvals (excluding COVID-19 devices), he said.

However, "There are existing gaps regarding the agency’s ability to effectively monitor drug and medical device supply chains, for instance," the FDA witnesses said in their combined written testimony.

"Similarly, the existing regulatory pathways do not provide adequate flexibility for medical device manufacturers to adopt best practices for development, deployment or continuous maintenance of AI/ML-enabled medical devices," they said.

Shuren said that, in order to activate small innovators in the space, the FDA needs access to large data sets and is looking at federated models and learning from the Veterans Affairs agency. 

To keep medical device cybersecurity in check with secure-by-design requirements, the FDA requires companies to design medical devices "in a way that allows it to be patchable," Shuren said.

Balancing LDT safety with innovation

Several committee members asked about approaches to the regulation of laboratory-developed tests. The final rule, released May 6, affirms the FDA's position that LDTs are in vitro diagnostic products, regulated as medical devices under the Federal Food, Drug, and Cosmetic Act.

Guthrie questioned barring hospitals' and patients; access to tests under "onerous requirements" under the new regime proposed by CDRH, and Rodgers described the ruling as "executive branch overreach" that would affect how providers diagnose and care for patients, in a previous statement.

"While the final rule is a slight improvement to the proposed rule, it will still increase costs and decrease access to diagnostics and medical tests that provide information crucial for doctors to treat their patients effectively," she said.

Congresswoman Diana DeGette, D-Colo., noted that LDTs now use complex software that "inform extreme life-implicating decisions." 

While the final LDT Rule maintains exceptions, as previous enforcement discretion elements wane over four years, it lists significant factors that could cause a diagnostic to be stalled in FDA review. Guthrie cited 664 clinical holds in 2017-2021 and FDA's policy of written-only responses to applicants – with no dialog with applicants.

"It isn't acceptable," Guthrie said.

Of note, lab and diagnostic tests that are altered with new technology – such as the addition of AI or ML – would prompt the need for premarket review and compliance with the FDA's quality system requirements under the final rule.

Troy Balderson, R-Ohio, asked about facilitating the approval of AI/ML products that FDA has said it cannot accommodate, and how FDA can change that.

Shuren said FDA has authorized 900 AI/MNL, but shares that concern. 

"Those particular functionalities are kind of the lower-level diagnostic," he explained. 

"When we move into generative AI, the current [regulatory] framework does not fit the purpose, really designed around hardware."

Shuren said that AI requires more of a post-market model, so that FDA can monitor what is happening, in order to reduce risk of bias.

He also said the number of what is going to be produced using AI/ML is going to "skyrocket," and the agency would never have the resources to review all of it.

"It is unrealistic to expect that," and regulating the technology requires a third-party certification model, he said. "But that would require help from Congress."

Dr. Miller-Meeks accused CDRH of stifling innovation with the 500-plus page rule and questioned why real-world evidence is sometimes used and sometimes ignored. 

In a previous answer, Shuren said that there are "problematic" tests out there that are not performing well for patients, and they may not get the right treatment. 

"Today, if you are a cancer patient, whether or not you get the right treatment depends more on the lab that you go to, than your tumor biology," he said. "Not acceptable." 

To evaluate models that underpin algorithms, the FDA has the expertise, but they don't have the infrastructure, Dr. Patricia Carvizzio, director of FDA's Center for Drug Evaluation and research, said, noting the high cost of reviewing the models.

"We're finding ways in the interim," but the FDA is looking at how to upscale for the future.

FDA asked about CHAI

Miller-Meeks also asked if the FDA would outsource certification to the Coalition of Healthcare AI, and said that it was not a diverse body. She noted that Google and Microsoft are founding members, while Mayo Clinic, which she said has more than 200 AI deployments, employs some of its leaders.

"It does not pass the smell test," she said, and shows "clear signs of attempt at regulatory capture."

Shuren answered that, while CDRH engages with CHAI as a federal liaison, as it also does with other AI industry coalitions, it does not engage the organization for application reviews. 

"We've told CHAI, too, that they need to have more representation in the medtech side." While his office might consider CHAI's work, he said, "they don't work for us, and we don't work for them."

Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org

Healthcare IT News is a HIMSS Media publication.

Want to get more stories like this one? Get daily news updates from Healthcare IT News.
Your subscription has been saved.
Something went wrong. Please try again.