Hacker News Reader

Comment by wcedmisten | Hacker News Reader

I recently did a writeup [1] on the 510(k) FDA clearance process (which cleared this device for market).

Basically this device was cleared through a DAG of other devices that were "substantially equivalent". I made a website to visualize these relationships, and any recalls that occurred in the parent devices. If anyone is curious about this particular device, see here [2]

[1] https://wcedmisten.fyi/post/medical-device-analysis/

[2] https://www.510k.fyi/devices/?id=K232380 (click on "Predicate ancestry graph")

View on HN

The “substantially equivalent” approval loophole is a textbook example of regulatory capture:

- approvals for version N + 1 are basically skipped, as is safety testing

- manufacturers get a liability shield

- for many classes of devices, ladder pulling implies that no one will ever get another v1 approved. (The current rules for that are impossible to meet, but the incumbent company is grandfathered in under the old rules).

The testing is NOT skipped. This is a misunderstanding of 510(k).

The device still has to go through testing, validation and verification. It is a very detailed process and follows IEC spec. The device also needs IEC 60601 testing by a third party lab. In this case at least 60601-2-24.

The 510(k) means it is not doing something new therapeutically therefore it doesn't need to go through PMA process which so much longer and more complex. In this instance they're saying a pump that uses electronics to monitor and deliver glucose already exists. Our device does the same thing therapeutically but we do it differently and here is all out docs showing the device is safe.

If they came up with a magical patch that used quantum chemtrail energy to align the shakras and thereby affect the patient's insulin levels, then they would need to go through PMA and show the therapy is safe and works in small clinical trials followed by larger trials before they can make a device that can be marketed.

It isnt capture. IT is the opposite. Capture is how first movers and established players raise the bar for others. the 510k process does the opposite. it means that the FDA has already reviewed devices of that nature for safety and efficacy, and you dont need to do a clinical trial, you just need to do the testing

to correct your points:

- No testing is skipped (just trials

- There is no liability shield.

I have no idea what you are talking about with ladder pulling. Do you have examples?

The testing to show "substantial equivalence" are not generally trivial. The "original" device needs to show safety, and effectiveness versus clinical endpoints. The follow up devices need to show safety and effectiveness versus appropriate measurements. Some degree of safety testing and/or analysis is generally always required for medical devices, regardless of if its a PMA or 510k.

For the specific case of blood glucose monitors, the 510k submission will require testing to show that your monitor is safe, and provides "at least as good" blood glucose monitoring compared to predicate devices. This allows the device manufacturer to avoid having to prove that blood glucose monitoring will improve clinical outcomes for patients with diabetes. Testing to show performance typically requires some degree of real world (ie, with real operators and patients) testing.

There are certainly be some cases where this gets abused, but overall I think it's a logical system.

Wow, awesome work! It's crazy to think about how the FDA itself probably can't answer a question like "how many predicates does device X have", but I guess also on brand with the idea that government is incompetent more often than not.

Trying to enact a policy that a limits the number/age of predicates in a 501(k) application is impossible if you have no way to find them!

>Wow, awesome work! It's crazy to think about how the FDA itself probably can't answer a question like "how many predicates does device X have"

Im not sure why you would think that. The answer would take 5 minutes to answer with the PDFs (which is where the parent post got the info).

For this particular device, there were 40 devices in the ancestry tree. While certainly feasible to traverse a graph of 40 PDFs and keep track of these relationships, it would probably take more than 5 minutes.

Additionally, not having the data in a structured format prevents seeing overall trends in the data.

I agree it would take longer to map them all. This is fine if you never need or want to map them. I've worked in this industry for 15 years and I have never had a need to do it.

The FDA already has information on product codes. here is the one for this device [1]. What is the useful trend and data one gets from grouping an iPhone app with a physical test strip for measuring glucose.

I think there is a lot of confusion in the general public about the 510k process is and what substantial equivalence means. Product drift is an intentional feature of the process, not a bug. 510k filing is not intended to be a single point control for safety and efficacy, but is implemented in context of other controls like medical device reporting and quality management.

Devices are primarily regulated at the device level, sometimes at the group level, but never at the lineage level. Im not sure what it would mean to trend and regulate as a lineage. Would that mean recalling a safe and effective product because of bad ancestor?

https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfTPLC/tp...

> DAG of other devices

Its well known you can infer the safety of a specification by assessing a small subset of its features. Quod erat demonstrandum. /s

But seriously, this is how backdoors are trivially integrated.

Reply on news.ycombinator.com