Free qtec qompakt webinar: PMCF – Clinical Data for Post-Market Clinical Follow-up

Question from our seminar participants:

Answer by Prof. Michael Imhoff:

“In MDCG 2020-5, there is a very detailed list of “equivalence” and a less detailed description of “similar devices”. Roughly speaking, the equivalent product must be practically identical to my own product. Especially when it comes to the topic of functional equivalence, discussions often arise with notified bodies. What is meant by this? Let’s take the example of two ventilators that are technically different but do the same thing for the patient. This means that the patient does not notice any difference between ventilator 1 and ventilator 2 because the performance data is identical in the end. Here we regularly encounter a need for discussion as to whether these two devices can then be considered as equivalent. With “similarity” it looks like this: As an example, let’s stick with the two ventilators. They are similar because they do the same thing, namely ventilate, and are used for the same indication. This means that I cannot conclude from the similar ventilators that the performance of my own device corresponds to the performance of the similar device. But I can use it to prove the general clinical benefit of ventilators, or as a benchmark device to compare the data of my own device with it. I can and should also use data from similar devices to gain further insights into device safety. For example, if I find safety problems with a similar product, I should check whether I may have similar sources of risk with my product. For me, “equivalent” means “practically identical”, “similar” means “they do the same thing but are technically the not same”.”

Question from our seminar participants:

“Regarding the calculation of the number of cases: How is it determined which rate of safety-related events (in the example: 1/1000) is acceptable?”

Answer by Prof. Michael Imhoff:

“In risk management, you have certain frequencies for the acceptance of residual risks. You use these for your estimates. For example, you accept a low health impairment in one in a thousand cases. However, if I see a risk acceptance of 1:1000,000 in risk management and my study can only statistically recognize a frequency of 1:1000, then I would question this as a reviewer of a notified body.

As far as the concrete case number estimation for a clinical study or a PMCF survey is concerned, you need assumptions, e.g. about the expected occurrence of an event or on the error range of the determination of the main target variable. Such a case number calculation is of course easier if you already have certain information on the expected frequencies or error ranges. This means that you have a certain idea of how widely the data can scatter. If you have no prior data, then you have no choice but to make a reasonably qualified estimate: “I expect (for whatever reason) the determination of my target variable to have a standard deviation of no more than X.” You then determine the planned number of cases, taking the desired power into account. You may be completely wrong with this estimate, but experience has shown that if you use the clinicians’ estimates of error ranges, your estimate should be in the right direction.

I haven’t actually seen a study being under-powered for such reasons. If you use estimated values because you couldn’t use prior data for the study, you rather run the risk of overestimating the number of cases in the study. This does not reduce the quality of the study, but it does increase the costs.”

Question from our seminar participants:

“If there are publications with a sufficient number of cases about off-label use, may a manufacturer extend the intended purpose/patient population based on this?”

Answer by Prof. Michael Imhoff:

“This is a discretionary decision by your Notified Body in Europe or the European Commission or the relevant Directorate. In the U.S., the FDA will tell you quite clearly whether this is feasible or not. In principle, you must of course consider all clinical data as part of your assessment procedure, including all data from off-label use. In the end, you will receive an answer to the question of whether you need a clinical trial or not. You then discuss this with the Notified Body, possibly with the Commission or (in the U.S.) with the FDA. If it turns out that all the studies were pushed by your sales organization, follow-up questions may arise. Otherwise, if you say that it is not your fault that the product is used off-label, but it seems to work well and there is enough clinical data, then it is worth a try. However, you must justify this approach.”

Question from our seminar participants:

“We had a case where a breast coil was used for an MRI device with the indication “female patients”. Now it is said that men can also have breast cancer. The intended purpose was to be extended accordingly, which led to problems. How could something like that be integrated into the intended purpose at a reasonable price?”

Answer by Prof. Michael Imhoff:

“A coil is an accessory for a scanner and it does nothing but improve the signal-to-noise ratio. Breast cancer in men is very rare, so it can be left as an off-label use. I would be relatively generous. Alternatively, you can conduct a survey of users and see how often the case occurs at all. Then you can check whether you can find a simple way to extend the intended purpose.”

Question from our seminar participants:

“We have never used surveys before and ask ourselves, why do doctors do this? A response rate of 30% isn‘t really high.”

Answer by Prof. Michael Imhoff:

“Various factors play a role in obtaining the highest possible response rate, such as

The length of the survey: the shorter, the better (ideally 5-10 minutes)
How much does the doctor enjoy working with the product?
Is a small donation offered for completing the survey?

If you have a stable sales organization or if you do direct sales in clinics, then the sales staff can ask directly on site whether there is interest. However, at the moment when I create surveys where managers have to obtain real case numbers or data, I expect a relevant effort from the respondents, which should then also be remunerated accordingly. This is no longer a compliance problem if there is work involved. We have had rather positive experiences with surveys so far.

Basically, however, you should always expect lower response rates and be happy if they turn out to be higher in the end.”

Question from our seminar participants:

“What tools do other companies use to effectively obtain feedback from product users?”

Answer by Prof. Michael Imhoff:

“We have a fundamental problem here when it comes to incidents involving medical devices. The reporting rates on the clinical side are suboptimal. This is also due to the fact that in the hospitals the healthcare professionals, especially physicians, do not have the capacity for such tasks, and the administrations say – not wrongly from their point of view – that this is not directly related to patient care.”

Question from our seminar participants:

“How do I deal with small amounts of clinical data in terms of significance/statistics?”

Answer by Prof. Michael Imhoff:

Question from our seminar participants:

“What significance do clinical data from your own product have without a comparison group?”

Answer by Prof. Michael Imhoff:

“That depends on the question I want to answer. If I want to say that the product has a better benefit than another product, then of course I have to compare it. But if I say I’m only looking at safety, then I look at the number of events in relation to the applications and compare that with my own threshold values from risk management. Then I don’t need any comparative data. So, it really depends on the question.”

Question from our seminar participants:

“How extensive does PMCF need to be?”

Answer by Prof. Michael Imhoff:

“That depends a lot on the product. For established products that have a low risk, you can keep the PMCF very lean. For newer products or products with a high risk, you need a comprehensive PMCF.”

Question from our seminar participants:

“How can the lifespan of absorbable medical devices be proven if an additional intervention (X-ray) is not proportionate.”

Answer by Prof. Michael Imhoff: