Regulation of devicesBMJ 2010; 341 doi: http://dx.doi.org/10.1136/bmj.c5730 (Published 03 November 2010) Cite this as: BMJ 2010;341:c5730
The development, approval, and surveillance of medical devices have some features in common with those for prescription drugs, as well as others that are even more challenging. Yet the regulatory traditions of the device world are strikingly different from those of drugs and often much more problematic. This is shown in the linked feature by Lenzer and Brownlee (doi:10.1136/bmj.c4753), which describes the history of a vagus nerve stimulator to treat epilepsy—a device with a worrying risk-benefit ratio⇓.1
It would be unthinkable for a drug company to go to the Food and Drug Administration (FDA) of the United States or the European Medicines Agency and say, in effect, “This new drug is much like an older product we sell, except that we have added a new amine group and modified one of the side chains. Apart from that it’s pretty close, so we won’t be doing any new clinical tests on it. When can we begin marketing?” But this is essentially what happens with many new medical devices when they are approved. Unlike drugs, a new device can be approved for use if it is “substantially equivalent” to an existing product.2 The industry argues that many of its products, such as defibrillators or pacemakers, represent incremental modifications of predecessors, so the efficacy and safety of the older products can be presumed to apply to the new version as well.2
The device industry has also persuaded its regulators that the usual evidence from randomised controlled trials required for drugs is not necessary.3 This perspective begins with the idea that it would not be ethical to take two patients with severe osteoarthritis of the hip (or even more starkly, a femoral fracture) and randomly allocate one to a sham operation, while the other gets the miracle of titanium. But such plausible arguments have morphed beyond their ethical justification, so that the standards of clinical trial assessment (and even the requirement for conducting trials in humans) are dramatically lower for devices than for drugs.4
These methodological concerns have combined with an increasingly powerful and assertive medical device industry and (in the US) a long standing crisis of organisation and leadership in the branch of the FDA that oversees these products. As a result, the standards for device approval and surveillance have fallen far below those for drugs, and even those that would be dictated by common sense.
A surprising and totally unnecessary administrative deficit on the clinical side has made the problem much worse. Whereas every prescription drug dispensed in most developed countries is recorded somewhere in a digitally retrievable code describing it in detail, no such requirement is uniformly applied to devices. As a result, when a patient receives an implantable pacemaker or defibrillator in most US settings, a researcher or regulator can readily identify all the drugs the person was taking before the procedure, but a detailed definition of the device itself is not routinely recorded in most clinical databases. Therefore, when the modified wiring of a pacemaker is found to cause potentially fatal short circuits (the electrical equivalent of a lethal new molecular side chain that was never clinically tested), it is hard to perform systematic assessment, notification, or recall. Even car manufacturers have better databases to identify who uses their products, as numerous communications from Toyota to its customers have made clear in the past year.
Surveillance of practice has been equally lax. For years, surgeons have installed stents to treat peripheral arterial disease. But most of the stents used in these procedures had been approved only to treat bile duct stenosis. Despite widespread use in treating peripheral arterial disease, these products were not subjected to regulatory review for this purpose, a subject of intense current litigation.5 In the vital area of postmarketing safety surveillance, voluntary spontaneous reports are still the main means of detecting problems with devices currently in use. This is a notoriously inadequate approach even when it is well managed, which it often is not. In the drug world, the withdrawal of rofecoxib (Vioxx) made it clear that adequate surveillance cannot be based merely on the passive receipt of spontaneous adverse event reports. The approach is even more problematic if device manufacturers can decide which serious side effects they choose to report. In contrast, for new drugs all serious adverse events must be reported by the manufacturer, regardless of their nature or presumed causality.
Fortunately, several administrative and scientific solutions exist. The new team running the FDA has changed the leadership of the branch that oversees devices, which is a welcome development. The agency needs to reconsider the dangerously minimal “substantially similar” standard for approval, and put better requirements in place for clinical review of new products, as was done for drugs in response to the thalidomide disaster of the early 1960s.
After the withdrawal of rofecoxib, Congress insisted that the FDA build an ambitious “sentinel” system to review the use of drugs and clinical experience of millions of patients on a near to real time basis. Similar programmes have existed for years in the United Kingdom (the general practice research database) and the Netherlands (Pharmo), among others. The increasingly complex world of medical devices needs to make the same transition, and it could take advantage of some of the same databases, if they were modified appropriately. Mandatory recording of the detailed specifications of installed devices in existing healthcare datasets will be a necessary step in this evolution.
In pharmacoepidemiology, methodological advances are helping to tackle the thorny problem of confounding by indication—the problem that a given disease (such as coronary artery disease or epilepsy) can cause the same adverse outcomes (myocardial infarction or sudden death, respectively) as the treatments used to manage it. Rigorous definition of all exposures and outcomes will be necessary even to attempt to define these associations correctly.
Organisationally, the disappointing history of device regulation reminds us that it is not wise to allow the safety of a medical intervention to be judged primarily by the people who approved it and the companies that manufacture it. Placing safety surveillance activities at a higher and more independent position within regulatory bodies would be one solution to this problem. Doing so would also make it possible to pool epidemiological expertise across several domains in which similar problems arise, such as drugs, biological agents, devices, and—in the case of the FDA—food borne illness too.
Complex medical devices are an increasingly prominent and vital component of the medical armamentarium. Doctors, patients, and policy makers are awakening to the need for greater rigor in the evaluation and surveillance of all healthcare interventions. As the world of prescription drugs emerges from its own difficult past in these respects, its recent experience holds important lessons that the world of medical devices should be required to assimilate as well.
Cite this as: BMJ 2010;341:c5730
Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declares: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
Provenance and peer review: Commissioned; not externally peer reviewed.