Note: This post appeared in abbreviated form as the Final Word article in the September/October 2019 issue of BI&T.
The history of regulation indicates that it is very difficult to design comprehensive frameworks to deal with new and emerging technologies in advance of them being introduced to clinical practice. This is largely because any new technology comes into use with a series of more or less easily identified risks, as well as a number of unknown risks.
In the language of Donald Rumsfeld, these are the “unknown unknowns.”
Viewed from 36,000 feet, regulation of medical devices and pharmaceuticals is essentially the same. How much evidence is required (i.e., identification of known and quantified risks and known and unquantified risks) to allow a product to be used on patients at the outset? Then, the conditions for early use need to be designed to ensure that both anticipated and unanticipated risks can be monitored in an effective way and that patient exposure to those risks is managed. This is the essence of effective postmarket surveillance, and it should confirm that assumptions tested premarket are borne out when the product is in general use.
It is important to understand that postmarket surveillance and vigilance are very different. All parties need to move away from a historical perception that just looking at adverse events constitutes compliance with a manufacturer’s postmarket surveillance obligations. It is likely that if the first signal that all is not well comes from vigilance reports, then the postmarket surveillance program has not been designed effectively.
Indeed, postmarket surveillance should and must evolve across the life cycle of the device and as experience and knowledge of risks evolve. This is particularly important if trials supporting market entry do not reflect the full gamut of potential applications of the technology and “real-world evidence” indicates that unanticipated risks (and benefits) are emerging.
The detail of what is required for individual categories of products will inevitably be very specific and determined by the nature of a given product. Existing structures and tools are increasingly challenged by very rapid technological evolution in the medical device arena, as well as by increasingly sophisticated combination products, which span devices, medications, and software.
It might be tempting to be very prescriptive about regulatory requirements and evidentiary needs, but inevitably, such an approach runs the danger of disproportionate hurdles for some products and their iterations and inadequate evidence for others.
This means that all involved in the regulatory process need to apply large amounts of common sense. It also means that increasingly, regulators will need to develop clear guidance around the limits of such interpretative activity. Category-specific guidance on needs for clinical evidence is high on the agenda for regulators across the world.
Are current regulatory paradigms fit for the future of health technology? The answer is both “yes” and “no.” Yes—we must build on existing tried-and-tested regulatory frameworks, but these need to evolve rapidly to accommodate the needs of technology that challenge these old models. And no—because current regulations were designed for mechanical engineering– and electrical engineering–based products made in factories with established quality management systems.
These models will need to flex to accommodate near-patient manufacturing, software and artificial intelligence (AI), genomics, and a host of previously unimagined technologies, along with their mode and location of manufacture. The current regulatory frameworks do not fully accommodate managing risk in these new areas, but starting with a blank sheet of paper to try and design a completely new set of regulatory frameworks would do little to advance safety in the application of these new technologies.
I believe that new regulatory frameworks will emerge, but they will need to evolve from existing frameworks and take elements with them. A proliferation of individual frameworks might prove very difficult to manage; therefore, I do not see this happening, except perhaps in the area of software and AI, where it seems that the difference from conventional medical technologies is so profound that it may require a framework of its own.
In the absence of any other appropriate starting point, medical device legislation is a good place to start. “Evolution not revolution” is my mantra, but evolution needs to be fast if environmental change is rapid.
John Wilkinson is director of devices at the Medicines and Healthcare products Regulatory Agency in London, UK