For discussion's sake - by showing that the device's nature & design are such that there's no reasonable doubt about it's safety (including unacceptable side effects) and performance, that would need proof from an actual clinical setting. I realise that this probably narrows down the applicability to simple, low-risk devices, but I think that this was the actual intent. If the device works in a complicated, harder-to-predict way, how can you show that it's safe and effective other than through actual, statistically valid use?
Yeap, but then again we go back to begin requiring clinical investigations for a lot of devices that are in the market today. And you are right, the intent seems to be this, but what I'm arguing is that applying the intent for all devices (in particular when using the new
MEDDEV) may seem a bit unfeasible, for several reasons.
I do, however, argue that yes, most manufacturers need to have clinical evidence on file. I also think most of them actually have it, but many are too lazy to collate it and analyse it properly, or are trying to save the costs of a properly done literature search and analysis. If a device is not novel or has been around for long, there must be some sort of cumulative evidence for its safety and performance.
My experience is different. Most actually do not have it (and it's worse with the MEDDEV makes it more difficult to use "equivalent" devices, which was the justification in the past) and this include all types of manufacturers I work with. And they are not trying to save on the cost. But anyway, maybe it's only I that am experiencing this.
If a device is not novel or has been around for long, there must be some sort of cumulative evidence for its safety and performance.
Sure, but not in "clinical data" as defined. They may have a lot of other data, for example, low count of adverse events, etc. But this is not clinical data. So if we say that these devices, which may not be low risk, simple devices, need clinical data, these other justifications do not count, we do need clinical data and, due to the lack of data, we need a clinical investigation.