Yale researchers discover loophole in FDA medical device regulation

Researchers at the Yale School of Medicine and Harvard Medical School found that a loophole in existing regulation has allowed manufacturers to acquire U.S. Food and Drug Administration approval for unsafe medical devices.


Stephanie Hu

1:50 am, Jan 26, 2023



U.S. Food and Drug Administration

A recent study led by researchers at the Yale School of Medicine and Harvard Medical School found that a loophole in existing regulation has allowed manufacturers to acquire U.S. Food and Drug Administration approval for unsafe medical devices. 

This work was led by Kushal Kadakia, first author and M.D. candidate at Harvard Medical School, and Harlan Krumholz ’80, senior author, Harold H. Hines, Jr. Professor of Medicine and director of the Center for Outcomes Research and Evaluation. Their study found empirical evidence that medical devices approved based on a previously-recalled device through the 510(k) regulatory pathway were significantly more likely to be subject to a Class I Recall, the FDA’s most severe designation for recalls.

“The 510(k) pathway does not require medical devices to undergo new testing as long as they can show they are substantially related to previous approved devices, known as predicates,” Kadakia said.

This pathway expedites the approval of medical devices that may only have minor changes from previously approved iterations and are being used for the same purpose. In fact, over 95 percent of new devices are cleared by the FDA through this pathway.

But due to a loophole in the regulation, the predicates themselves may not actually be safe for human use. 

“The way the law is written, if the FDA pulled it off the market, it can’t be used as a predicate, but if the company pulled it off the market, you retain the ability to reintroduce a new one that is substantially equivalent and still be used for that unsafe purpose,” Krumholz said. 

The study focused on medical devices that were subject to a Class I Recall. This sort of recall is issued when a medical device has a reasonable probability of causing severe adverse health consequences up to and including death.

Previous studies had provided case studies showing harm caused by devices approved using recalled predicates. Kadakia worked on two such studies of a catheter and sleep apnea device that were later subject to Class I Recalls. This new study is unique, however, in its scope.

“We were able to go across several years and identify all the devices that had these recalls, instead of picking out one or two,” Krumholz said. “We were able to look at a comprehensive group and give a more representative view.” 

This approach was made possible by recent advances in machine learning and data science. Because the FDA’s database only contains decision letters, which list the reasoning behind an authorization, it can be difficult to figure out what devices have been authorized using a specific device as a predicate. Without the use of new computational tools, it would have been time-consuming to map the lineages of medical devices. However, the researchers were able to construct these lineages in partnership with an AI company and then manually confirm the AI database’s results.

The researchers found a 6.4 times increase in recall rates for medical devices approved using recalled predicates when compared to non-recalled predicates. Given that each device can have tens of thousands of units and are used throughout the medical process, these recalls can have widespread effects.

The Safety of Untested and New Devices Act of 2012 was a previous attempt to rectify this issue, but failed to secure enough votes. The researchers hope this novel study may reinvigorate the United States Congress to at least begin discussion of the 510(k) pathway again.

“The recalled predicate loophole is not an unknown quantity in Washington,” Kadakia said. “We have now provided empirical evidence in a systematic way of how this loophole is being used to cause harm.” 

The study authors also acknowledge that more work can be done using these new computational methods. 

“We limited it to a one generation analysis, but it would be interesting to look at the children of children of recalled predicates and so on,” said César Caraballo, a postdoctoral associate at Yale School of Medicine. 

Krumholz hopes that more evidence would strengthen Congress’s ability to enact wise and empirically sound legislation. This is especially critical as medical devices receive far less research attention than drugs because they are embedded throughout the medical process instead of at the point of care, Kadakia explained. 

“If we were able to add unique device identifiers to claim forms, we could quantify the amount of spending that was authorized through the predicate recall loophole,” Kadakia said. “We could also determine if the reasons for the new recalls and the recalls of the predicates are similar.”

In the fiscal year 2022, 149 medical device products were subject to Class I recalls.

Leave a Comment