Alexandra: 0:15

Medical devices are more prevalent in our lives than we think. Ranging from simple bandages to MRI machines, every single one of us has used a medical device at some point in our lives. However, the pharmacovigilance of medical devices is anything but simple. Beyond the medical risks, there are technological and engineering risks associated with the use of devices. My name is Alexandra Coutinho, and this is Drug Safety Matters, a podcast by Uppsala Monitoring Centre where we explore current issues in pharmacovigilance and patient safety. Joining me on this two-part episode is Omar Aimer, the president of ISOP and the co-lead of the ISOP Medical Device Safety Special Interest Group. In part one, we discuss what medical devices are, what makes the pharmacovigilance of medical devices more complicated than medicines and vaccines, and the regulatory efforts currently in place to ensure medical device safety. I hope you enjoy listening.

Alexandra: 1:17

Hi Omar and welcome to Drug Safety Matters. Thank you for taking the time to talk today on what is becoming an increasingly important topic in pharmacovigilance, and that is medical device safety. How are you doing today?

Omar: 1:31

Good, good. Thank you. Thank you for having me.

Alexandra: 1:34

It's lovely to have you as well, and thank you for making the time to talk with us. So, to help our listeners better understand what a medical device is and how prevalent they are in society, can you give some examples ranging from the simplest to the most complex devices we might encounter in healthcare and our everyday lives?

Omar: 1:55

Yes, certainly. Medical devices range from simple to highly sophisticated devices, especially nowadays. Maybe I can start with the definition of a medical device. So the definition of the IMDRF, which is the organisation of the harmonisation for medical devices, that we'll talk about later in this podcast. A medical device is any instrument, apparatus or software used for diagnosing, treating, mitigating or preventing a disease or abnormal physical condition in humans. This definition includes a wide range of items from simple tools like tongue depressors and bandages to complex equipment like pacemakers, X-ray machines, and in vitro diagnostic tests. A key distinction is that medical devices do not achieve their primary action through pharmacological or metabolic means, which is the difference with the drugs or medicines and vaccines and other health products. So on the simpler end, we have bandages, thermometers, and contact lenses. In the middle range, there are blood pressure monitors, glucose meters, and infusion pumps. But at the most complex level, we find MRI machines, pacemakers, as I mentioned, and robotic surgical systems. All those medical devices play vital roles in modern healthcare and daily life for patients.

Alexandra: 3:36

I didn't even know that bandages qualifies as a medical device. That's quite interesting. They are very inherent in even my everyday life. When I was thinking about a medical device, I was thinking about a pacemaker, those progesterone sort of injections that you get. And I didn't think that they were as common as you say they are. So, what is the status of safety monitoring for these medical devices? And what are the key differences between monitoring devices versus traditional pharmaceuticals and vaccines, given how extremely complex medical devices can be?

Omar: 4:07

Great question. This is really the difference with the traditional PV or pharmacovigilance. We all know in the pharmacovigilance and the safety space that the globalization of pharmacovigilance was triggered by the Thalidomide tragedy in the 1960s. The medical device safety was triggered later in the 2010s after the PIP tragedy in France, with a problem related to the breast implants with very poor quality that was manufactured by a company in the south of France. That triggered all this globalisation of medical device safety and also the updates on the regulations reinforcing the post-market surveillance. The current status of safety monitoring for medical devices is maturing globally and even implementation. Many regulators have strengthened device vigilance. I mentioned the postmarket surveillance, there's mandatory reporting, there's public adverse event databases that were implemented, but the adoption and the capability vary widely between the regions. There's really an advance in EU with the EU MDR and IVDR, but there's a growing phases on proactive surveillance. Regulators and industry increasingly use registries, real-world data, like for drugs and medicines and vaccines, the device-specific post-market clinical follow-up, and the signal detection tools, rather than relying only on spontaneous reports. And also the regulatory convergence is underway, or the global collaboration between regulators. There's some regional initiatives and systems, such as the Eudamed in Europe, some other national databases, like one in the US, the GUD ID or the MAUDE database, are pushing for the harmonisation. But gaps remain in data standards and cross-border sharing, and also the support for the low and middle-income countries. There's an increase to push more this post-marketing surveillance for medical devices.

Alexandra: 6:32

I understand then speaking on the status of safety monitoring, but are you saying that there aren't really that many differences between medical device monitoring and that for pharmaceuticals and vaccines?

Omar: 6:45

Yes, there's some differences. Like also the seriousness criteria, not the same. There's the engineering part in the medical devices. The use of the medical devices is more complex than the use of the traditional medicines. There's differences, of course. There's the notion of the users. It can be the patients, it can be the healthcare professionals, there's the human factors as well on medical devices. Yeah, there are devices, and the action is not the same as for the medicines and other health products.

Alexandra: 7:21

Okay. I understand. So just as you said, there seem to be several converging factors from technological advances to changing patient demographics. I guess that's what you meant by then the human factor element of medical device safety monitoring that make medical device safety monitoring more important than ever. From your perspective, what are the key drivers behind this increased urgency to monitor the safety of medical devices?

Omar: 7:48

The safety monitoring for medical devices, called also medical device vigilance, has advanced significantly, but still varies across the region. As I said, the manufacturers and regulators now rely on post-market surveillance, the real-world data, and incident reporting. And here we have not only the adverse event reporting or AE reporting, but we have also the incident reporting to detect and manage safety issues once the devices are in use or are marketed or approved. Compared with the pharmaceuticals and vaccines, medical device monitoring is more complex because devices can malfunction, can degrade, or interact with users and software in ways that drugs cannot. It's not the same interaction. And unlike the medicines, whose safety largely depends on biological effects, device safety often involves engineering performance, usability, and cybersecurity as well, because some applications and applications are considered as medical devices, and we will talk about this later. Additionally, the software updates or component changes can alter a device risk profile, requiring ongoing dynamic monitoring throughout its life cycle, and all the health application that we are using in our smartwatches or smartphones that measures some health signs are considered as medical devices. And there's data, there's updates, there's AI on it as well that need to be monitored because it's evolving and there's change in the device risk profile.

Alexandra: 9:46

Wow, yeah. And I guess in what you're saying, you mentioned, for example, earlier post-market surveillance. I'm also starting to understand how important post-market surveillance then is even after a medical device has been approved. And now I can understand why, considering you're talking about how it evolves as a person is using it. You recently published a paper in the journal Drug Safety discussing ongoing challenges in medical device safety vigilance, despite previous calls for improvement. What do you see as the greatest barriers preventing more effective monitoring systems?

Omar: 10:20

Yeah, great question. Yeah, despite the progress in the medical device manufacturing and post-market surveillance, several barriers still limit effective medical device safety monitoring. The most significant are fragmented data systems where post-market information is scattered across manufacturers, hospitals, and regulators, making it difficult to detect signals early for medical devices. Another challenge is the lack of standardised reporting and the coding for device-related incidents, which hampers data aggregation and analysis, which is similar with drugs but more complex for medical devices. In addition, the limited interoperability and the real-world data integration, particularly for devices connected to software or AI that I mentioned earlier, slow down the signal detection. Finally, there's often insufficient feedback between users, manufacturers, and regulators, delaying corrective actions to the malfunctions, to the incidents that occur or that happened with these devices. Strengthening the collaboration, the data sharing, and global harmonisation are really a key to overcoming these barriers.

Alexandra: 11:49

You also mentioned AI, and your paper also discusses how AI-enabled and other smart devices like smartwatches are creating new safety challenges. Can you explain what performance drift means in AI medical devices and why this poses unique monitoring problems?

Omar: 12:08

Yeah, so the performance drift refers to the gradual change in how an AI-enabled medical device performs over time, as the data it encounters in real-world settings differs from the data it was trained on. The ChatGPT like today that you are using, and the ChatGPT that you will use tomorrow is not the same. If you ask the same question, you will have a different response. It's really based on all the information that the AI was trained on. For example, the changes in patient populations, in this example, for medical devices, clinical practices or data quality can cause an algorithm's accuracy to decline without anyone immediately noticing. It's really evolving. The AI was trained on day one on information, and this information is evolving in day one to 100, and then there's this performance drift. This poses unique monitoring challenges because, unlike the traditional devices, AI systems can evolve or adapt. Sometimes even automatically. The continuous vigilance is therefore needed to ensure that an artificial intelligence model remains safe, reliable, and unbiased throughout its use. It requires continuous vigilance. The traditional one-time validation isn't enough. Instead, we need ongoing performance monitoring and revalidation to detect and correct the drift before it affects patient safety. So the continuous monitoring and the revalidation are key to prevent performance drift.

Alexandra: 14:09

Okay. Just to make sure that I understood you correctly, your paper mentioned that when AI is used in medical devices, it's trained on a fixed set of data. So are you basically saying that because it's trained on this fixed set of data, this is not what happens in reality. In reality, demographics, our information is constantly evolving. So whatever the AI has been trained on is no longer relevant to the evolving real case scenario. And that's what this performance drift is?

Omar: 14:39

Yes, exactly. You know, a medical doctor fresh after the graduation starts his experience as a medical doctor, and his medical judgment right after his graduation is not the same as his medical judgment 20 years after. Because of the set of information that they can have during their career and experience, and of course, evolution of the science. So, yeah, the set of data that was used to train this AI on day one will be evolving, the science evolving. So the diagnosis can be drifted or the performance can be drifted and should be monitored, should be revalidated again.

Alexandra: 15:29

Okay, yeah, understood. Changing tact a little bit. Um you also spoke a little bit about drug device combination products in the paper, like insulin pumps. And these drug device combination products present what you call dual safety challenges. How should regulatory systems adapt to better monitor these hybrid products?

Omar: 15:50

Drug device combination products, such as you mentioned, insulin pumps or inhalers, pose dual safety challenges because they involve both pharmacological effects and mechanical or software performance. A failure can stem from either component or from their interaction. It can be from the medical device, it can be only from the medicine or the drug, but it can be also from both. There's plenty of examples like insulin pumps. If the insulin pump is malfunctioning, if the pen gives half of the insulin, then the patient will have hyperglycemia because the patient doesn't receive a sufficient dose of insulin. This is because of the medical device. But sometimes the problem is with the insulin and not with the device, and the patient will have hyperglycemia or hypoglycemia, and to address this, the regulatory systems need to adopt integrated oversight that bridges traditional drug vigilance and device vigilance frameworks. Of course, the trainings of the pharmacovigilance professionals is also a key to detect those failures or those adverse events or incidents. It depends on the problem that occurred. But this includes harmonised reporting systems, shared signal detection databases, and cross-disciplinary review teams that combine clinical, engineering, and data science expertise. In essence, regulators should treat these hybrids or combination products not as separate entities but as interdependent systems, ensuring continuous monitoring of both the drug safety profile and the device's real-world functionality throughout their life cycle. So we need really to oversee the adverse event or the incident by assessing both the device functionality or functioning and the drug safety profile. It's more complex than the traditional PV for drugs only.

Alexandra: 18:16

Yeah, I mean I think that your answer moves us really neatly onto the next question that speaks about harmonisation, which seems to be especially important when we're combining pharmaceuticals with engineered medical devices. The paper emphasises the critical role of the International Medical Device Regulators Forum, so the IMDRF that you mentioned before. These are a voluntary group of medical device regulators from across the world. And you mentioned how important they are in harmonising medical device regulation efforts. Can you explain what the medical device single audit program is and how it's helping to improve global safety standards?

Omar: 18:54

The ICH is known for the pharmaceutical world as the organisation for the harmonisation for drugs or for medicines. In the medical device world, we haven't a similar organisation than the ICH with a scope in harmonising efforts. And then we have the IMDRF, which is the International Medical Device Regulators Forum, as you mentioned in your question. So the medical device single audit program is an initiative led by this organisation that allows a single regulatory audit of a medical device manufacturers quality management system to satisfy the requirements of multiple participating countries simultaneously. So yeah, for the manufacturers, it is very important because it goes smoothly and eases the barriers of the different regulations in different regions, especially for the global organisations. By streamlining audits across different jurisdictions such as the US, Canada, Australia, Brazil, and Japan, this medical device single audit program reduces the application, saves resources, and promotes consistent enforcement of quality and safety standards worldwide. This harmonised approach helps ensure that medical devices meet high safety and performance criteria everywhere they are marketed in the countries that I mentioned, ultimately improving global patient safety and facilitating more efficient regulatory oversight.

Alexandra: 20:48

That's all for part one, but we'll be back soon with the second part of the episode where we shift our focus to the other side of medical device pharmacovigilance, the challenges of reporting a medical device adverse event. If you want to know more about this topic, check out the episode show notes for useful links. And if you'd like more pharmacovigilance stories, visit our news site, Uppsala Reports, at UppsalaReports.org to stay up to date with news, research and trends in the field. If you enjoy this podcast, subscribe to it in your favorite player so you won't miss an episode and spread the word so other listeners can find us. Uppsala Monitoring Centre is on Facebook, LinkedIn, X, and Blue Sky, and we'd love to hear from you. Send us comments or suggestions for the show or send in questions for our guests next time we open up for that. You can also visit our website to learn more about what we do to promote safer use of medicines and vaccines for everyone everywhere. For Drug Safety Matters, I'm Alexandra Coutinho. I'd like to thank my colleague Federica Santoro for post production support, and of course, you for tuning in. Till next time.