PharmedOut, a project of the Georgetown University Medical Center, had its third annual conference a couple weeks ago in DC. The meeting was called “Missing The Target: When Practitioners Harm More Than Heal,” and it addressed some of the most important issues in health policy—issues that have been disturbingly absent from the public discussion. While the news media and members of Congress have spent the last several years discussing insurance coverage (or, more often, what the most recent vacuous coverage-related talking point means for the political horse race), issues of patient safety have been left on the sidelines. The PharmedOut meeting was a welcome opportunity to return to focusing on patients.
One of the most compelling moments came during Dr. Amy Friedman’s talk. Dr. Friedman is a transplant surgeon, and frequently speaks to groups of surgeons. She asks them a number of questions during these presentations—her favorite of which is “True or false: All medical devices used in the US have been proven safe and effective.” A large majority of the surgeons she talks to know that the answer is “true.”
The trouble is, of course, it’s not.
I can understand members of the public getting that wrong--after all, the Food and Drug Administration is responsible for making sure that all foods, cosmetics, pharmaceuticals, and medical devices are safe, and that the latter two are effective. But it's inexcusable for doctors to be so ignorant of the weaknesses in our regulatory system.The fact is, there are gaping holes in device approval processes, to the extent that many medical devices in use today have never been tested for effectiveness.
When the FDA started regulating medical devices in 1976, the agency established rules that assumed any device already in use was safe and effective (unless it was already known not to be), and those devices continued to be sold. That grandfather clause included a number of basic and obviously safe devices, like tongue depressors and scissors, but it also covered higher-tech devices like the implantable pacemaker (invented 1958).
The FDA also waives testing requirements for "low or moderate risk” devices that are "substantially equivalent” to devices already on the market, in what’s known as the 510(k) process. For those devices, the device maker doesn't have to get approval from the FDA to sell it--they just have to give the FDA notification before they sell it. That means minor modifications to existing devices don’t require new clinical trials. When used intelligently, the rule is entirely appropriate--there's no need to redo clinical trials for small tweaks to products that can't affect the safety or effectiveness of the device. As the FDA has applied it, though, the 510(k) rule has become a giant loophole, and the vast majority (over 90%) of newly approved devices are approved through that pathway.
The problem is simple if you think of medical devices as undergoing evolution: each new device is slightly different from its parent (known as the "predicate" device). Once it's on the market, the modified device can be used as a predicate for another modification. Over time, compounding small changes can get you a tool or implant that bears no resemblance to the original device that actually went through clinical trials. Worse, some of the original predicates are those devices the FDA grandfathered in, and may never have gone through clinical trials.
The holes in the approval process have caused real harm. Take the example of metal-on-metal artificial hips: DePuy ASR hips were approved based on substantial equivalence to another hip with a metal head and a ceramic or polyethylene socket joint, so they were never tested prior to market. It was only after thousands of people received the hip that it became clear that the metal-on-metal joint was scraping off bits of cobalt-chromium alloy into patients’ bodies—and worse, the hips were failing early and often. If clinical trials had been conducted, thousands of patients might have been saved the pain and hassle of a failed hip and an extra operation to fix it.
There’s a better way to do device approvals—a middle ground between requiring randomized trials for every minute change in a device’s design and letting device manufacturers run completely free. We should be tracking what happens to patients after they receive new or experimental devices. We should develop a national, automated database that registers each patient who receives a recently-approved device, and monitor those patients for device-related complications for several years. During that time, all patients receiving the new device should be informed that their device is still semi-experimental, and that they are part of a Phase IV clinical trial to find out more about what the device actually does. If that information makes patients want to use older, better-understood devices, that's not a reason to claim regulations are holding back progress in medicine--it's giving patients complete information, and letting them choose the care they want. Forcing patients to use new, unproven devices (by failing to tell them about the experimental status) is no more ethical than denying patients new treatments in favor of old.
Tracking outcomes in a large-scale database works: a similar registry was a key part of the Australian research that discovered problems with DePuy hips. Reporting doesn't don’t have to be intrusive or an administrative burden, either. Wal-Mart and Amazon collect data on every transaction, every day, to better understand their customers and manage their inventory; customers hardly even notice. Even the data analysis can be largely automatic.
Tracking new, untested devices is a no-brainer. There are legitimate privacy concerns with such a database, but large datasets with medical, personal, and financial information are routinely scrubbed so researchers can’t find private information about any individual. There’s no excuse for continuing to let Wal-Mart do better at understanding what affects its sales than researchers and doctors do at understanding how to make patients healthy.