Medical Devices: Gaps, Tensions, and Conflicts in the FDA Approval Process:: Medical Devices
Author: Richard A. Deyo, MD, MPH, Departments of Medicine and Health Services and the Center for Cost and Outcomes Research, University of Washington, Seattle
The FDA's approach to approving medical devices differs substantially from the approach to drugs, being in some ways both more complex and less stringent. The FDA's authority over devices dates only to 1976. Device legislation was a response, in part, to public outcry over some well-publicized device failures. The most prominent was the Dalkon Shieldan intrauterine contraceptive device associated with serious infections. In contrast, the FDA's authority over drugs dates to 1938, although it existed in weaker form starting in 1906.
With few exceptions, given the timing of the FDA's authority, devices introduced before 1976 were never required to undergo rigorous evaluation of safety and efficacy. With the huge volume of "things" that suddenly fell under its purview, the FDA had to prioritize its resources and efforts.
One way of prioritizing was to focus first on safety. Evaluation of effectiveness, in many cases, was reduced to engineering performance: does the device hold up under its intended uses, does it deliver an electric current as advertised? The potential benefits for relieving pain, improving function, or ameliorating disease did not generally have to be demonstrated.
Another way of prioritizing was to assign categories of risk associated with the devices. Rubber gloves seemed less risky than cardiac pacemakers, for example. So the agency assigned devices to 1 of 3 levels of scrutiny. Class I devices have low risk; oversight, performed mainly by industry itself, is to maintain high manufacturing quality standards, assure proper labeling, and prevent adulteration. Latex gloves are an example.
At the other extreme, class III devices are the highest risk. These include many implantable devices, things that are life-supporting, and diagnostic and treatment devices that pose substantial risk. Artificial heart valves and electrical catheters for ablating arrhythmogenic foci in the heart are examples. This class also includes any new technology that the FDA does not recognize or understand. New components or materials, for example, may suggest to FDA that it should perform a more formal evaluation. In general, these devices require a "premarket approval," including data on performance in people (not just animals), extensive safety information, and extensive data on effectiveness. This evaluation comes closest to that required of drugs. In fact, Dr. Kessler says, these applications "look a lot like a drug applications: big stacks of paper. They almost always require clinical dataalmost always. And they often require randomized trials. Not always, but often" (L. Kessler, personal communication). These devices are often expensive and sometimes controversial because of their costs.
Class II devices are perhaps the most interesting. They comprise an intermediate group, generally requiring only performance standards. Examples would be biopsy forceps, surgical lasers, and some hip prostheses. The performance standards focus on the engineering characteristics of the device: does it deliver an electrical stimulus if it claims to, and is it in a safe range? Is it made of noncorrosive materials? Most of these devices get approved by the "510(k)" mechanism. The 510(k) approval requires demonstrating "substantial equivalence" to a device marketed before 1976. "And," says Kessler, "the products that have been pushed through 510(k) are astonishing" (L. Kessler, personal communication).
Kessler points out, "For the first 5 to 10 years after 1976, this approach made sense. But in 2001, 25 years after the Medical Device Amendment, does it make sense? There was a lot of stuff on the market that wasn't necessarily great in 1975why would you put it back on the market now?" (L. Kessler, personal communication). The new device need not prove superiority to the older productjust functional equivalence. If a company wants to tout a new device as a breakthrough, why would it claim substantial equivalence to something 25 years old?
The reason is that the 510(k) process is easier and cheaper than seeking a premarket approval. The 510(k) process usually does not require clinical research. In the mid-1990s, a 510(k) application on average required 3 months for approval, and about $13 million. A premarket approval required, on average, about a year and $36 million. Both are modest compared with new drug approvals. The process by which the agency decides if something is "equivalent enough" to be approved by the 501(k) mechanism is subjective.
Because pre-1976 devices were not subject to any rigorous tests of clinical effectiveness, a newly approved device may be equivalent to something that has little or no therapeutic value. Doctors, patients, and payers therefore often have little ability to judge the value of new devices. As an example, the FDA still receives 510(k) applications for intermittent positive pressure breathing machines. Yet a thorough review by the federal Agency for Health Care Policy and Research found that these devices offer no important benefits.
How much do manufacturers take advantage of the easier 510(k) approach? Since 1976, nearly 98% of new devices entering the market in class II or III have been approved through the 510(k) process. In 2002, the FDA reported 41 premarket approvals and 3708 approvals through the 510(k) process.