The FDA is known as having the gold standard for ensuring the safety and effectiveness of medical products, but those standards differ substantially for devices compared to prescription drugs. The standard for medical devices is “a reasonable assurance of safety and effectiveness,” and that is not the same as “safe and effective.” It is not clearly defined, and as a result, those standards vary widely.
Prescription drugs are approved based on scientific evidence that is almost always based on clinical trials – studies of patients with the disease or illness that the drug is intended to treat. Almost all these drugs and biologics are compared to a placebo control group, because patients sometimes do well with a treatment that they believe is effective, even if it is just a placebo pill. Having a placebo control group is considered an important standard for a clinical trial, to reduce bias. However, fewer than 5% of medical devices regulated by the FDA are tested in clinical trials. Approximately 95% go through an easier pathway that is called the 510(k) process. That is based on evidence from the company that its device is “substantially equivalent” to another medical device that is already on the market. There is no requirement that the device it is similar to has ever been proven safe and effective in a clinical trial, and in fact, usually they aren’t.
Medicare almost always pays for prescription drugs that are approved by the FDA for a particular indication, such as a breast cancer drug for adults. But, it does not automatically pay for all medical devices, because it is understood that the standards are lower for getting medical devices on the market. Our research has shown that most medical devices do not need to have clear scientific evidence that they are proven safe or effective, and often there is no public information about any evidence that they are safe or effective [1]. Since the Congress is considering requiring Medicare to pay for all breakthrough devices, we will consider the evidence for patients ages 65 and older.
Breakthrough medical devices are considered by the FDA to be high priority, and the selection criteria require them to be more effective than alternatives for diseases or conditions that can be life-threatening and irreversibly debilitating. Each device selected must also meet at least one of 4 criteria, such as representing a breakthrough technology, there are no FDA-approved or cleared alternatives, has significant advantages over other cleared or approved options, or is beneficial to patients. FDA designated more than 1100 breakthrough devices through early 2025, based on preliminary information that it can meet the required criteria, and the devices then have to provide evidence that they should be available to market, but do not have to prove that they satisfy the criteria for being a breakthrough device. That resulted in 160 breakthrough medical devices cleared for market, many of which would not satisfy the criteria initially required to be designated as breakthrough.
Of those 160 breakthrough devices, this article will focus on the 112 that are therapeutic devices (for treatments, not diagnostic tests).
Previous research based on 75 therapeutic breakthrough devices has found that 67 (89%) underwent premarket testing, usually based on short-term follow-up (median of 6 months for implanted devices). What follows is our data on 510(k) devices, since they are the devices least likely to be proven safe and effective even if they are designated as breakthrough devices.
One of the 50 510(k) devices is for infants and toddlers only, so we will focus on the 49 breakthrough 510(k) devices for adults. Only 8 of the 49 treatment/therapeutic devices (16%) were in clinical trials that were listed in ClinicalTrials.gov. Two other devices seemed to have clinical trials but no trials were listed for those devices in clinicaltrials.gov. Listing in clinicaltrials.gov is required by FDA law, with the exception of very preliminary studies, so it would be odd to require Medicare to pay for any device that did not obey that law.
Of those 8 devices, one intentionally excluded any patients over 63 years of age from the clinical trials. That leaves only 7 (14%) of the 510(k) breakthrough devices that satisfy the criteria of having a clinical trial that might have been studied on patients ages 65+. Although the law requires that the results of clinical trials on clinicaltrials.gov be included online, information about the age of patients in these studies is not available on 6 of the 7 devices. There is publicly available info on only one of these devices that clearly states it includes patients 65+.
The clinical trials above are all based on the new 510(k) device. However, there are 17 510(k) devices (35%) that have some kind of clinical data (not described as a clinical trial or listed on ClinicalTrials.gov) and many of those include data from the predicate (previously cleared device) rather than the new breakthrough device. In addition, there are 10 more devices that provide nonclinical data that is also related to their predicate. In total, that means that more than half (27/50 breakthrough 510(k) devices) provide information based partly or primarily on the predicate, not the new device, whether that information is clinical information (often based on a few patients or on a previous study of the predicate that was not described) or animal or other nonclinical data.
Since 510(k) devices are “substantially equivalent” to devices on the market, it is not surprising that many breakthrough 510k devices are not obviously superior to other devices on the market. Here’s one example, which is especially concerning because it is for spinal fusion surgery, which can be dangerous under the best of circumstances.
Carlsmed aprevo™ Transforaminal IBF (patient-specific lumbar interbody fusion cages).
The Carlsmed aprevo™ device is a 3D-printed, patient-specific fusion cage, which is a spinal implant used to help treat severe spinal deformities in adults who have not improved with standard care. It is custom-made for each patient, using CT scans to better match a patient’s anatomy, with the goal of improving fit and surgical outcomes. The FDA cleared it through the 510(k) pathway based on bench and engineering tests (not trials on humans), comparing it to similar devices already on the market. Despite addressing a serious condition, no studies comparing patient outcomes using this device compared with other devices were submitted to demonstrate it was more effective, even though superior effectiveness is a required criterion of breakthrough devices. This is an example of a breakthrough device that reached the market while lacking any studies of humans (or animals) that proved it met the requirement of being better than the many other spinal fusion cages that are cleared by the FDA – or even if it was as safe or as effective as other spinal fusion cages. The cost is not publicly available, but we expect a custom-made implant to cost more, and CMS pays approximately $20,000 more when a spinal implant is “new technology.”
Here is an example of a 510(k) device that was recalled 3 times and is still on the market.
Ominipod Insulin Management System
The Omnipod insulin delivery system is a device worn on the arm to calculate and control the amount of insulin provided to people with diabetes. It has been recalled three times due to problems with accuracy, delivering too much or too little insulin. In a 2024 Class I recall (the highest risk recall), the Android app misread certain decimal entries (e.g.,”0.2 units was misinterpreted as “2.0”), which would deliver ten times more insulin than intended and could result in life-threatening hypoglycemia. In the previous (2022) Class I recall, the Omnipod DASH handheld controller faced a battery overheating and fire hazard. A separate Class II (moderate risk) recall involved an error message in the Omnipod 5 Android app that temporarily blocked phone-based insulin dosing, causing therapy delays. These design-related software and hardware flaws affected insulin delivery in dangerous ways. We are glad that these recalls resulted in fixing the problems, but meanwhile, how many patients were harmed? When serious design problems like this are not identified before a device goes on the market, it can be fatal, which is why this device should have been tested on humans before allowing it to be sold.
- Zuckerman D.M., Brown P. & Das A. (2014) Lack of Publicly Available Scientific Evidence on the Safety and Effectiveness of Implanted Medical Devices, JAMA Internal Medicine, 174(11): 1781-1787.