There's been a great deal of interest in the FDA review of the 510(k) processes. One of the more thoughtful discussions of the recently published FDA report comes from Barry Sands at WMDO. Barry pointed out the "unusual level of openness and honesty" in the report, then went on to say:
I actually believe the review hit upon the majority of issues that are of concern and some of the recommendations are sound. However, there appears to be an underlying belief that this regulatory process should produce a risk free regulatory decision with regard to the premarket clearance/approval decision. I believe that one needs to understand that the regulatory process itself has risks that cannot be mitigated to zero. Just like every single medical device available on the market today possess risk, so does the regulatory decision to allow or not allow a medical device to enter and remain on the market.
Sands points out some specific concerns about knowledge base and inconsistency of interpretation, but then returns to a discussion on risk that goes to the essence of the different approaches of the US FDA compared with the "GHTF model" regulators such as Europe, AUstralia and Canada.
The core issue raised is whether FDA has a mandate to consider device "benefit" rather than "effectiveness" (Sands argues not). The former is a measurement of clinical outcome, the latter a measure of engineering performance.
In the non-US, GHTF world, it's quite clear that the onus on the manufacturer is to demonstrate effectiveness, the benefit is more a matter for clinician decision and choice based on the performance claims and evidence. That's why clinical trials to support CE mark tend to be simpler, faster and cheaper than those to support US approval (particularly PMA). THe European trials in essence focus on confirmation of real world performance of the device, which can be done more quickly than an efficacy study which may need to follow long term clinical outcomes.
It remains to be seen where the FDA review will lead, but Sands raiises a very important issue. he put it very well:
If we attempt to drive the associated risk of the regulatory decision to zero, we may very well have a process that cannot make a decision.