Rethinking medical device regulation

Complications with Poly Implant Prosthese (PIP) breast implants ¹ and Metal on Metal (MoM) hip implants ² reflect systemic failings with the current regulation of medical devices. Yet, these two cases highlight only a fraction of the burgeoning increase in medical device safety alerts ³ and problems with device recalls, and are leading to a rethink of the systems for regulatory approval in both Europe and the USA. ⁴

Therefore, having an understanding of medical device regulation is now an important requirement for doctors and healthcare professionals alike. To aid this, French-Mowatt and colleagues summarize the current medical device regulation in Europe, ⁵ outlining the current requirements for CE regulation. Outside of the European Union, Susan Lamph describes the regulatory processes across different countries and the lack of harmonization with leads to wide variation in pre-market data requirements. ⁶

Analysis of medical-device recalls in the UK and the USA, and the device-regulation process, reveals the increasing nature of the problems. From 2006 to 2010, the UK regulator, the MHRA issued 2,124 manufacturer field safety notices, an increase of 1,220% over this five-year period. ³ In the USA the number of recalls for moderate or high-risk devices more than doubled between 2007 to 2011. ⁷ In addition, many recalled medical devices in the USA were originally cleared using a less stringent process called 510(k), or even more problematic, recalled devices were considered to be so low-risk they were exempt from regulatory review in the first place. This situation reflects a very low ‘bar’ currently for evidence requirements to gain regulatory approval, even for high-risk devices. For instance, of 78 high-risk cardiovascular devices approved through the more stringent Pre-Market Approval (PMA) regulatory process in the USA, fewer than one-third had undergone randomized trials and only half of the trials overall involved controls. ⁸

Whilst new drugs require at least randomized controlled trials to gain regulatory approval, for medical devices even under the more stringent PMA approval process, only one controlled trial (not necessarily randomized trial) is required. However, an even more worrying issue with device regulation in both the EU and US is the use of ‘substantially equivalent’ in evidence submissions for regulatory purposes. The problems with using ‘equivalence’ in the device approval process can be traced back several decades. In 1976, in the USA many devices were already on the market, so a less burdensome alternative to PMA known as 510(k) provision was approved. The 510(k) pathway did not require clinical trials; the manufacturer was only required to demonstrate a device was ‘substantially equivalent’ to another device already on the market. The problem now is that the definition of equivalence is interpreted so loosely that the FDA admits they need to ‘clarify the meaning of “substantial equivalence.”’ ¹⁰

The predicate of equivalence is also used within the European Union (EU) regulatory system for device regulation. There are three European Directives related to device regulation. ^11–13 These directives, which lead to CE marking and access to the European market, state the extent and nature of clinical data required for approval. ⁵ Problems occur because even for implantable devices, the scrutiny of evidence at the outset is left to private organizations known as Notified Bodies; ³ and secondly, clinical data required for the equivalent route can involve as little as ‘a critical evaluation of the relevant scientific literature currently available relating to the safety, performance, design characteristics and intended purpose of the device’. The use of equivalence is therefore left to the manufacturer and the Notified Bodies to determine, without any outside scrutiny of the decision making process centrally or within each EU country.

Even for the more stringent PMA process, there are profound differences in evidence requirements between the US and EU. For example, EU approval of a ‘GuardWire’ developed by PercuSurge for use during angioplasty, required a 22 patient study with no control group. Yet, in the USA, FDA regulators required an 800 patient multicentre randomized controlled trial. ¹⁴

Perhaps what is even more concerning than the device recalls and high profile cases (such as the MoM hips and PIP implants), is that many medical device problems go unnoticed. For example, women participating in a breast cancer study were left with hundreds of tiny particles of the heavy metal tungsten in their breast tissue due to a faulty device cleared under the 510(k) processes. ⁹ Similarly, recalled device notices often go unheeded. In 2006 the maker of a surgical clip, the Hem-o-lok issued an urgent recall notice warning surgeons to stop using the clips on living kidney donors. However, three years later a surgeon used one of the clips to tie off a 29-year-old male's renal artery during an operation in which he donated a kidney to his wife. He bled to death twelve hours later when the clip malfunctioned. ^10–15 Currently we have limited ability to trace most patients in whom medical devices have been used (or implanted), so when problems or recalls occur, it can be impossible to know the magnitude of the problem.

However, it seems as though the tide is turning in terms of regulatory requirements. The American system is coming under increased scrutiny with calls for the removal of the 510(k) process. The influential Institute of Medicine has recommended the FDA do away with the 510(k) approval process and replace it ‘with an integrated premarket and post-market regulatory framework that effectively provides a reasonable assurance of safety and effectiveness throughout the device life cycle.’ ¹⁶ It is possible that all implantable devices in particular will require PMA approval and thus clinical trial data in the future. However, more stringent regulations are unlikely to be passed into law in the US without a substantial battle with the medical device industry.

Sprange and Clift's analysis of manufacturers' submission challenges, to the NICE medical technology program, in the supplement edition J R Soc Med 2012;105 (S1) reveals there are significant issues in relation to basic and general research skills that need to be addressed amongst manufacturers. ¹⁷ In addition, interviews with manufacturers highlight the current status quo: ‘pharmaceutical and medical technologies were also considered very different by manufacturers.’ As such, the widespread belief is that devices do not require the same level of evidence as drugs to gain access to a market and be used in clinical practice. Sprange's study highlights the need for education and research ‘tools,’ which will facilitate higher quality evidence submissions for approval in the future.

The European Directive on medical devices will be revised later this year. The European Commission has stated it will use this opportunity to strengthen existing legislation, particularly provisions relating to market surveillance, vigilance and the functioning of notified bodies. In the UK, the House of Commons Science and Technology Committee plans to examine whether current legislation and regulations on safety and efficacy of medical implants are fit for purpose. ¹⁸

Failures of medical devices cause harm and cost money. More stringent requirements to provide evidence from clinical trials for the efficacy and safety medical devices before they are approved should therefore be welcomed by patients, clinicians and the medical device industry. Evidence for new devices must also be open to scrutiny by patients in individual countries, as well as healthcare providers and researchers. The potential risk of a new device should match the type of evidence required prior to approval for use in clinical settings. Without these changes to current systems, it is likely we will continue to see substantial complications arising from faulty devices.

DECLARATIONS

Competing interests

None declared

Funding

The authors received no funding

Ethical approval

Not required

Guarantor

Carl Heneghan

Contributorship

Both CH and MT contributed to the ideas, drafted the manuscript and approved the final version

Acknowledgments

None