A Little Learning about Testing Medical Devices
Friday, February 18, 2011
Recent calls for tighter clinical requirements for medical devices should themselves be recalled. Such requirements would not greatly increase public safety; they would hamper innovation.
Our esteemed colleague and friend Jack Calfee died Wednesday evening, February 16, of a surprise heart attack, just two weeks short of his 70th birthday. He was struck while working at his desk, polishing the essay that follows, written with his research assistant, Gabriel Sudduth.
During his 16 years as an AEI resident scholar, Jack’s research, books, and innumerable papers made immense contributions to our understanding of innovation, intellectual property, the economics of information, and government regulation. He was a great champion of medical innovation and its contributions to health and welfare—and a great critic of efforts, by regulatory agencies and liability judgments, to control and suppress innovation, medical practice, and the free flow of information to consumers. Jack was also a great economist: he was always eager to put his views to the empirical test, and his work brimmed with original insights and subtle interpretations.
Jack’s first AEI publication, appearing in Regulation magazine in 1986, documented the Federal Trade Commission’s successful efforts to shield consumers from information about the health risks of cigarette smoking—in the name of “truth in advertising.” In this final brief piece, he returns to the health hazards of well-meaning regulatory policies that are uninformed about the dynamics of markets and innovation. — Nick Schulz, Editor-in-Chief
The Archives of Internal Medicine, an American Medical Association journal, recently posted two articles on medical device safety and the Food and Drug Administration’s (FDA’s) device approval process. They add to a staccato of recent news reports about medical devices, which are a world unto themselves, ranging in complexity and intrusiveness from stethoscopes to artificial hearts. The article by Diana Zuckerman, Paul Brown, and Steven Nissen presented data on all the most serious type of FDA recalls (Class I) from 2005 to 2009—113 devices in all. Device recalls are safety alerts that include necessary action steps including returning devices to manufacturers, watching for potential complications or software upgrades, and halting use pending changes. Most of the recalls (80 total) involved devices approved through what is known as the 510(k) process rather than the more rigorous pre-market approval (PMA) process. To gain marketing clearance from the FDA in the 510(k) pathway, manufacturers must present significant engineering, testing, and sometimes preexisting clinical data to demonstrate that a new device is similar to existing devices. As they clear this pathway, the vast majority of devices do not go through clinical trials—a very different situation from that of new drugs, which almost always run through at least two trials.
Unlike drugs, devices change almost constantly—anything from software tweaks to better wiring insulation.
Disturbed by these facts, Zuckerman, Brown, and Nissen call for the FDA to require all high-risk devices (classified as Class 3 by the FDA) to pass through the more rigorous, drawn-out PMA route and, moreover, to expand the definition of “high-risk” to include any devices whose failure could potentially cause severe problems. Requiring devices to go through the PMA process would include more than just clinical trials; it would also bring stringent manufacturing controls and inspections, post-market surveillance requirements, and a quick product removal process. An invited “commentary” by Rita Redberg and Sanket Dhruva largely agreed with Zuckerman, Brown, and Nissen. The New York Times, Wall Street Journal, and Washington Post all printed stories that recited these basic points while also quoting industry sources who said the situation was not nearly as worrisome as the Archives suggested.
The Archives article authors provided a remarkably limited perspective, as did others such as Dhruva, Redberg, and Lisa Bero, and John Somberg, et al., who have called for tighter clinical requirements for medical devices, with drug trials as models. The assumption is that tightening requirements for many of the most important medical devices would greatly increase public safety without overly hampering innovation. Such calls typically pay scant attention to the vast disparities between devices and drugs. Unlike drugs, devices change almost constantly—anything from software tweaks to better wiring insulation. As manufacturers upgrade their technology and compete with others who produce similar items, calling an altered device “new” is often a matter of judgment. In any case, most “new” medical devices are very similar to preexisting devices but contain better technology to improve overall success, durability, and cost.
The assumption is that tightening requirements for many of the most important medical devices would greatly increase public safety without overly hampering innovation.
When Congress passed the Medical Device Amendments in 1976 and other laws in the early 1990s to add pre-market device clearance to the FDA’s mandate, it established the 510(k) approval mechanism precisely because everyone knew that many of the thousands of devices already on the market would need to be improved, and that a lot of new devices would be extremely similar to existing ones. There was no reason to treat all these as novel devices subject to burdensome requirements for new clinical testing. Legislators and the FDA also knew that some device modifications are best studied in the laboratory or only in animals, rather than in patients, because the crucial variable might be software, speed, or electrical performance.
In addition to ignoring how the device market operates, the argument for tighter clinical requirements also often ignores the 510(k) device safety record. Of 19,000 devices approved in the 510(k) process from 2005 to 2009, only the 80 devices examined in the Archives study were subject to Class I recalls. Because many device variables are adjustable, some recalls are quickly addressed. Recalls in the Archives study included software upgrades, electrical component changes, and the removal of a button cover. A customer could perform the latter action. Customers can also sometimes perform software upgrades and other changes themselves. Most recalls, however, require devices to be returned to manufacturers for replacement. Often, problems requiring a recall appear only after years of marketing and would not be caught in pre-marketing clinical trials, a point made by one manufacturer in the Wall Street Journal story. Again, Class I recalls are very few compared to the thousands of unrecalled devices cleared through the 510(k) process.
Congress established the 510(k) approval mechanism precisely because everyone knew that many of the thousands of devices already on the market would need to be improved, and that a lot of new devices would be extremely similar to existing ones.
The Archives articles also state that the FDA defies the law when it runs Class 3 (highest-risk) devices through 510(k). This is untrue. The FDA must write regulations in order to move a class of devices into the PMA mechanism. It has chosen not to do so for many device classes, leaving the choice of approval route subject to various factors including, of course, safety. The FDA announced in 2009, however, that it was forging ahead with the necessary regulation to force all Class III devices through the PMA process. Whether that is a good idea is far from clear. Evidence of systematic safety problems in the U.S. device market due to FDA laxity is lacking, and is certainly not provided by the Archives studies. The PMA process requires far greater investments of time and money than the 510(k) process, resulting in devices that are more expensive and, more importantly, much slower to reach patients. Though Zuckerman and her colleagues do not advocate putting all 510(k) devices through clinical trials, heeding their call to expand the definition of “high-risk” beyond Class III devices (effectively reclassifying some Class II devices), and to push the resulting devices into PMA, would probably leave many patients worse off.
Finally, what about Canada and the European Union? Do they provide the extra measure of protection that the FDA fails to provide? On the contrary, they do not impose anything like the FDA’s current requirements, not to mention the slower, more intensive, and more expensive approval system that the authors think the FDA needs. As the FDA and others rethink the 510(k) device approval process, they would do well to pay more attention to experience abroad.
John E. Calfee was a scholar at AEI since 1995 and a regular contributor to THE AMERICAN. Gabriel Sudduth is a research assistant at AEI.
FURTHER READING: Calfee previously exposed “Junk Science and the Anti-Vaccine Fraud,” discussed “Why the AARP Drug Price Reports Are Misleading or Worse,” and revealed “Something Old, Something New: Biotech’s Enormous Potential.” He outlined “Six Ways Not to Reform Healthcare,” considered “Drug Development in the Balance,” and explained why “Limiting Drug Prices Means Limiting Future Cures.”
Image by Rob Green/Bergman Group.