Risky Business: Expert Tells How Medical Devices Get Cleared in U.S.
For people with a type of abnormal heart rhythm known as ventricular arrhythmia, implantable cardioverter-defibrillators can be a real life saver.
These little electronic devices track heart rhythms and spot heartbeats that are off. When that happens, electrical shocks correct the rhythm, ensuring the heart keeps pumping on pace.
But what if these devices – or their components – fail?
It’s happened. And the consequences can be deadly.
In the early 2000s, for example, some implantable defibrillators were malfunctioning because of problems with newly marketed wires – or leads – that connected patients’ hearts to the devices. First, a recall was issued in 2007 for models of Medtronic Sprint Fidelis leads. And then, another brand of leads produced by Abbott Labs, known as Riata leads, were recalled several years later. But both those recalls came only after hundreds of thousands of patients worldwide had the devices already inside their bodies, hundreds of malfunctions had occurred and several patients had died.
Such a problem with medical devices critical to people’s health would seem an aberration in a developed nation known for its rigorous standards for medical care. But those recalls made up just a tiny chapter in the big story of medical devices that end up either not working as claimed or actually causing harm.
In fact, between 2002 and 2016, the U.S. saw 487 “high-risk” recalls, according to a report published earlier this month in the Journal of the American Medical Association by Jonathan J. Darrow, an assistant professor of medicine at Harvard, and medical doctors Jerry Avorn and Aaron S. Kesselheim, who are affiliated with Brigham and Women’s Hospital in Boston. As in the case of the leads, a number of those recalls came years, sometimes many years, after the products were OK’d by the Food and Drug Administration.
“The average patient expects that if something is authorized by the FDA, then it’s got the FDA’s watchful eye and one assumes that it’s pretty safe most of the time, that it can be used without worrying about it,” Darrow told TheStreet. “But the reality is that regulators don’t know everything. Industry doesn’t know everything. And there are risks.”
Jonathan J. Darrow, an assistant professor of medicine at Harvard
To understand how so many recalls can become normal under the nation’s health care regulatory system, it’s necessary to understand some history of federal legislation and policy around medical devices. Much of the story of modern device regulation begins in 1976, when legislation was introduced that started a device classification system based on risk. Over the decades, that regulatory system has become more and more complicated and new pathways to the market have been added.
But how could high-risk products like those implantable defibrillator leads get by regulators, despite ending up with fatal flaws?
Part of the reason can likely be traced back to FDA application processes that allow for reviews that sometimes don’t require clinical testing. That’s because the devices are believed to be sufficiently similar to products already on the market, like those leads. These roads to the market include the 510(k) pathway and a type of “supplemental” review process, known as a pre-market approval supplement. In essence, through these pathways, even moderate-or-high-risk products can be deemed essentially like their predecessors, but with incremental changes tacked on.
But, as seen with the leads, sometimes those changes can lead to big problems.
“Even if each change is individually small, an accumulation of changes over multiple product generations could result in large changes over time that are significant to patient health. There have been times when that’s happened – where the changes appeared to be small but were later discovered to meaningfully impact patient safety or efficacy,” said Darrow.
He noted that a 2008 study had shown that the Medtronic Sprint Fidelis leads had a change in width that had been made on the basis of bench testing without new human data.
Darrow’s JAMA paper also shows that only a tiny portion of all devices on the market get there through the strict process known as the pre-market approval process. Most of them come through the 510(k) pathways. Just how many? Between 1987 and 2020, a median number of 3,404 products annually came to market that way. By contrast, a median number of 32 devices each year won the more stringent pre-market approval.
Darrow, who is an affiliate faculty member at Harvard’s Petrie Flom Center and its Center for Bioethics, as well as an associate professor of law at Bentley University, spoke to TheStreet by phone recently about the findings in his paper. The following has been edited for clarity and brevity.
TheStreet: Coming away from this, I have to bet that very few people realize how complicated the FDA’s approval and authorization process can be for medical devices – and, in some cases, how little evidence of efficacy is needed compared to the approval process for, say, new drugs.
Darrow: One of the most important things that patients should understand is that there is no minimum amount of efficacy that is required – even for drugs – and the standards are generally even more relaxed for devices. The standards that are in place are not so much efficacy standards, but evidence standards. That means you have to have a certain amount of evidence and that evidence has to have a certain amount of reliability. It’s a very flexible standard, and the FDA has substantial discretion when determining if that standard has been met.
Devices are subject to risk-based evidence standards. This means that the highest risk medical devices tend to require – on average – the most robust evidence. This is called the pre-market approval pathway, or PMA. A very small share of devices, however, come through the pre-market approval pathway. Most come through the 510(k) pathway or are exempt from review.
The important thing patients should know is that even if the FDA has been provided with extremely robust evidence that supports the effectiveness of the device, the device might not be that effective. The magnitude of efficacy — how much the device will improve the patient’s condition—isn’t the main focus of the FDA’s analysis. Even if it is clearly characterized, which it might not be, efficacy magnitude is often not communicated to patients. That is unfortunate, because this information is essential to patient decision-making.
TheStreet: Talk a little about that 510(k) pathway…
Darrow: The way the 510(k) submission process works is that the submitter has to identify some predicate device that the proposed device is put forth as being substantially equivalent to. In some cases, that predicate device – on which the 510(k) submission is based – may have also been reviewed through the 510(k) pathway. So, you can have this series of linked authorizations under the 510(k) pathway that go back many product generations. Each product generation may reflect a small change, but over time, those changes can accumulate and yield a product that is quite different from the original, pre-1976 product. There is a concern that with this long-term evolution, larger changes that are significant for patients could accumulate over time. …
The 510(k) process was probably implemented in 1976, the year the Medical Device Amendments were enacted, in part because without it, incumbent manufacturers could continue to market their products, but any new entrants that wanted to compete would have to submit a greater amount of evidence to enter the market. This could create a monopoly-type situation that could yield higher prices, which isn’t a good thing for consumers.
But the other effect of the 510(k) pathway is that, all these years later, we have many devices on the market that have been determined to be substantially equivalent to preexisting devices, but if those preexisting devices were never evaluated for safety and efficacy, then there is potentially a substantial share of currently marketed devices for which we really don’t know how effective they are or even whether they are effective at all. Similar questions remain unanswered about safety.
The problem with imposing elevated standards, however, is that it could substantially increase costs, and if made retroactive, could lead to a mass exodus of products from the market. For new devices, if you are required to run large clinical trials before marketing, society likely would not get the volume of new devices that currently are introduced every year. The impact of a reduced number of devices is difficult to evaluate, and depends on the value of those devices, which, as I said, in many cases we don’t know. Maybe we don’t need thousands of new devices every year. This is not a question that we were able to answer in our paper. Either way, imposing heightened evidence requirements could change the device landscape substantially. Not only would doing so impose a large cost on industry, but it would also impose an enormous burden on the FDA, which would have to spend much more time evaluating each new product or product variation.
TheStreet: That gets us to recalls. Could you discuss that process?
Darrow: The FDA has authority to initiate removal of products from the market if they are not safe or effective. Normally, the way recalls work is they are made voluntarily by the manufacturer. Manufacturers act voluntarily in part because they know that the FDA could order a recall if it needed to, and generally they want to avoid antagonizing the agency. Manufacturers also act voluntarily, because a failure to act is not good for business. If a company discovers something about its product that could potentially harm a patient, the company will want to do whatever it can to avoid the harm before it occurs.
Recalls do not always involve removal. Sometimes a manufacturer that becomes aware of a problem could work with the FDA and health care providers to ensure that patients are more frequently monitored, for example, to see if the problem is emerging in that particular patient. Depending on the device, it may be possible to repair the device on site, or relabel it with different instructions or warnings.
It can sometimes take time after a device is on the market to learn about potential problems. There has been an increased focus on collecting evidence after product approval, and there are increased surveillance requirements. There are also post-market study requirements that can be, and are, imposed. That means that in some cases, new information is being collected based on surveillance of patients in routine clinical care. Patients may not be aware that their data are being used by regulators to make regulatory decisions. Even though patient data are being used, the patient is not formally a trial subject, so the protections associated with clinical trials don’t apply. The line between patients and clinical trial subjects is increasingly blurred. Of course, it is only sensible to make use of all the data that are available, so regulators can’t be faulted for wanting to extract value from ordinary patient encounters, but I don’t think patients necessarily realize how their data is being used and the uncertainties that remain for the devices that are already on the market.
TheStreet: What about the fee structure. Someone may look at the way in which companies appear to pay fees to get their devices approved and think, “Oh, they pay a lot of money and then they get their devices authorized – sometimes only meeting an arguably low bar” – is that a concern? Is there concern that that money is an influence.
Darrow: I think there is a concern. It’s not clear why you would ever want industry to pay fees to the agency that regulates that industry. Imagine if polluters paid the Environmental Protection Agency for its enforcement activities. It really creates a conflict of interest. Even though there have been measures taken to separate the collection of funds from the activities of the FDA personnel, in terms of approval and enforcement, I do think it is problematic. It would be much better if the FDA were funded 100% from Congressional appropriations. Either way, the funding is the same. It’s just whether it comes from taxpayers directly — or indirectly from taxpayers when they pay their insurance premiums that cover the medical devices. It’s not like we’re saving any money doing it this way; it’s just that we’re creating a potential conflict of interest – and it’s gotten worse over time because the amount of fees that are paid to the FDA have dramatically increased.
