I'm sorry, but this is just patently false. The FDA requires extensive V&V testing, especially for medical devices that have a significant chance to kill the people that it's used on. That, and you need to have a very sophisticated quality management system (QMS) in place that tracks all the changes you make during your design and test process, links up source requirements to passing test reports in an auditable way, tracks all of your manufacturing information (including which batches were tested using what procedures, and the results), all of which can be randomly audited at any time, no matter where the facility is located on the world.

You may be referring to the 510k (pre-existing predicate device) pathway, which has been abused by companies to leapfrog and skip _some_ of the more stringent test requirements, but they still need to prove similarities, and that requires real data. However, in the vast majority of cases, these devices are things like masks, IV tubing sets, etc, not complex life saving equipment, which gets an enormous amount of scrutiny. If anything, it's the medical device companies that are at fault for lobbying Congress to allow abuse of 510k pathways, and Congress for allowing these things to happen.

> I'm sorry, but this is just patently false. [..] The FDA requires extensive V&V testing [..] and you need to have a very sophisticated quality management system (QMS)

Paperwork, they require paperwork.

It’s the same issues that plague FCC and CE, the amount of tricks and handwaving you can get approval for makes a lot of this (self performed/self directed) testing completely useless.

There’s a whole market of “expert consultants” whose job is gaming the tests and finding loopholes, and as long as tests are conducted in this manner, they will exist.

That's a little reductive. Companies get shut down all the time when their quality management is not up to snuff (for a high profile example of this, look at CMS' shutdown of the Theranos labs in 2015). The system generates the paperwork, and the data enclosed is generated by the processes you have to put in place to get approval in the first place.

Like I said, the world of predicate devices is a different ball game, but you're not going to get by with hand waving on a device that can kill people.

> Companies get shut down all the time when their quality management is not up to snuff.

That’s not a good measure for the performance of the regulation. In fact I’d like to see the numbers on the exact reasons for those shut downs, because I’m willing to bet they’re related to bad classification and other technicalities.

If the FDA were to be a test, you’d want it to have good discrimination at a low cost.

Anyone who has been involved in the process knows the cost is enormous, and scandals like Theranos (and good but extremely late good cases) show that discrimination is not that good to begin with.

I have been involved in the process, my company has a de-novo device (with an issued IDE), and I can agree that it is expensive and time consuming, but what their requirements result in is ultimately a better product, better processes, and dramatically reduced risk.

These regulations are written in blood. They're difficult to comply with because if you invite poor engineering you will get poor engineering and dangerous devices.