Little pre- and post-market testing for most recalled CV devices

Most devices recalled by the FDA came to market under the 510(k) procedure, which does not require new clinical or safety data.

Most cardiovascular devices subject to a Class I recall by the U.S. Food and Drug Administration entered the market without adequately rigorous premarket clinical trials, a new analysis shows.

Even when clinical trials were conducted before approval, they were typically of lower quality, and most studies were nonrandomized, unblinded clinical trials that used surrogate endpoints without active control groups, researchers reported in a study published Sept. 16, 2024, in Annals of Internal Medicine. Post-market surveillance was also generally quite weak.

“Medical devices generally have very diverse testing requirements, but we showed that for cardiac devices, which are generally well-tested, those that had safety issues shortly after authorization tended to have poor premarket evidence and no postmarket evidence requirements at all,” senior investigator Kushal T. Kadakia, MSc (Harvard Medical School, Boston, MA), told TCTMD.

Over the 10-year period, there were 137 Class I recalls – the most serious, potentially resulting in injury or death – involving 157 unique cardiac devices, including automated external defibrillators, Intra-aortic pumps and control systems and implantable cardioverter defibrillators (ICDs) are the most commonly recalled devices. A total of 71.3% of recalled cardiovascular devices entered the market through the FDA’s 510(k) pathway. The remaining recalled devices (28.7%) were approved through the Premarket Approval (PMA) and PMA Supplement processes.

Dr. Steven Nissen (Cleveland Clinic, Ohio), who has also studied medical device recalls and the FDA approval process, said the 510(k) pathway has long been problematic.

“Medical device regulations are not rigorous,” he told TCTMD. “A lot of devices are approved under the 510(k) rule, which says a device can be approved if it’s similar to a previously approved device. A lot of these devices are approved that way, and that means they don’t go through rigorous testing. When that happens, mistakes are made, and ultimately patients get hurt.”

Major Recalls of Medium Risk Devices

The 510(k) process, which is typically faster than the more rigorous PMA process, is designed to support iterative innovation, with new premarket clinical trials not required if the product’s indication and technology are similar to an already approved “predictive” device — moderate-risk class II devices such as diagnostic catheters or intra-aortic pumps.

This path has been criticized in the past, however, with the Institutes of Medicine once recommending its abandonment in favor of developing a new approval process for Class II devices.

Higher-risk Class III devices, such as pacemakers, ICDs, stents, and bioprosthetic valves, are approved through the PMA pathway and require clinical evidence demonstrating safety and effectiveness. However, a new indication for use or a change in labeling for a device may be approved through the supplement pathway, a process that does not require supporting clinical evidence.

These devices will continue to be part of cardiac care because they are so important and offer tremendous benefits to patients. Kushal T. Kadakia

From 2013 to 2022, 137 recalls affected an average of 7,649 devices available for clinical use. The most common reasons for a recall were device design, process control, and component design or selection. Of the 83 recalls that included safety information, the median number of incidents per recall was 13.5 complaints. The four recalls involved 12 to 26 deaths.

All recalled products were classified as Class II intermediate risk devices (61.8%) or Class III high risk devices (38.2%). The median time between manufacturer initiation of recall and FDA public notification was 1.4 months. The recommended action in 57.7% of cases was to discontinue further use.

“Because our study is focused on class I device recalls, you would expect these major recalls to be the highest risk devices,” Kadakia said. “The 510(k) pathway is for low- to intermediate-risk devices. The fact that 70% of our sample, the majority, were authorized under the 510(k) pathway rather than the PMA pathway suggests that perhaps some of this risk stratification is inappropriate. The devices that we call intermediate- to low-risk devices that we allow into 510K: maybe that’s not the right level of risk designation.”

Of the 30 devices with premarket clinical trials, seven were approved under the 510(k) pathway, 17 as PMAs, and six as PMA supplements. A total of 35 studies supported these 30 devices, and while almost all (94.3%) were prospective, most used a nonrandomized design and were not blinded. Twelve studies used active controls as a comparator, while four used objective performance criteria or targets. In addition, most studies (79.4%) used surrogate markers as an endpoint. The median follow-up time in these premarket studies was 6 months.

“When we have evidence requirements, we should make sure that those evidence requirements are robust and based on good-quality evidence,” Kadakia said. “In our study, when studies are done with these devices in the premarket setting, often they are not the best studies. Not only are they not randomized, sometimes they don’t have control groups at all. Many of them have really short follow-up periods, which is concerning, especially if the device is intended for lifelong use.”

Without extensive premarket data to demonstrate safety and effectiveness, “there will be recalls, and they could be serious,” Nissen said. “I think we need a change in the regulations.”

Nissen said that advocates of the status quo claim that tightening the 510(k) pathway will stifle innovation, but he doesn’t believe that. He cited as an example the FDA’s 2008 decision to require cardiovascular outcomes studies for every new diabetes drug, based on a study he led showing an increased risk of heart attacks with rosiglitazone. Critics of that “extremely controversial” requirement said it would stifle development of diabetes drugs, but it did the opposite.

“GLP-1 agonists, SGLT2 inhibitors — since then we have done the most amazing and effective studies that have changed the practice of medicine,” Nissen said.

After-sales settings

In terms of post-marketing surveillance, only 22 of the withdrawn devices were required to complete post-marketing studies. Slightly less than half (48.9%) of the PMA-approved devices had post-marketing study requirements. Of the 28 post-marketing studies for which information was available, 20 were completed by September 2023. Importantly, 20 of the devices with a post-marketing study completion order had a class I withdrawal before study completion.

Kadakia said that if the purpose of the 510(k) pathway is to enable iterative innovation, it is essential to understand what happens to devices in the real world after approval. If regulators give manufacturers flexibility in premarket evidence requirements, there must be robust requirements in the postmarket setting.

Mistakes happen, and ultimately patients get hurt. Steven Nissen

The FDA has proposed changes to the 510(k) pathway, issuing new draft guidance on the type of evidence expected for devices, recommendations for using clinical data as part of an application, and best practices for selecting an appropriate predicate device. The draft guidance is currently open for public comment, but if adopted, it is expected to tighten the evidence requirements for a 510(k) application.

“Devices will continue to be a part of cardiac care because they are so important and offer tremendous benefits to patients,” Kadakia said. “But we need to really strengthen the evidence-generating apparatus for these devices to make sure that we have sufficient evidence throughout the device lifecycle about both the benefits of the devices and their potential risks, which sometimes only become apparent as we learn more about the devices in use and in clinical practice.”

Nissen said an approval path that falls somewhere between a 510(k) and a PMA would be a good idea. “Something that says it’s similar to a previous device but needs at least some safety and efficacy data to be approved,” he suggested.