When Diabetes Devices Fail

Interesting article, and illuminating for me. Although I have an inherent skepticism about the medical industry, it seems I might have been a bit too trusting when it came to diabetes device manufacturers, and too often dismissive of complaints of failure.

Use this link to sign up for device notifications, and just to note, it asks for your interest areas after you enter your email and subscribe:


VERY interesting article, James.

Unfortunately, the MAUDE database isn’t very intuitive or user friendly. It’s also limited to US events, so it’s not comprehensive.

When I was choosing my first pump 3 years ago, I spent nearly 6 weeks reading through the reports I found. Not many people are willing to read those reports, or could comprehend the language.

I’m surprised that this is enlightening for the previous posters, though. I consider you both to be exceptionally well-informed. What part exactly was surprising? Did you not know there was a database of events related to medical equipment, or just not that it was publicly available and searchable? Is it just the volume of significant events that occur.

Honestly, the public has misplaced trust in the FDA. They assume that the FDA stamp on something means it’s proven safe, and there’s very little about establishing safety at all. That’s why drugs all come with a mile long list of side effects, usually ending with something like “anal hemorrhaging, and/or death”. You have to establish that something is generally safe, but aside from that, it’s just about establishing your marketing with impressive clinical trials, paying the government it’s due, and internal lobbying power.

Heck, you can flat out buy an FDA stamp to put on your marketing. A device can be “FDA listed” with little to no scrutiny whatsoever, so long as your check clears. You think the general population knows the difference between “FDA listed”, “FDA cleared”, and “FDA approved”? Heck no! Their eyes glaze over and they forego any further questioning the moment they see the “FDA”.

1 Like

As I stated in the original post, I am already very skeptical of device manufacturers and pharmaceuticals, and although not stated here, I’ve written about FDA ‘capture’, as well as industry research arms biased to justify the product. That said, I was still surprised about the laxity. Maybe I assumed the long times to market meant the devices were being vetted properly, but it seems less so than I believed, and obviously there are few after-market controls.

In the end, this isn’t a complete change for me, just that I need to set the bar a little lower, take a slightly dimmer view of the industry, and keep a more watchful eye.


I’m not sure I got everything, since there is a long list of announcement subscriptions but I signed up for the following, which is the heading and the bullet is the option under the heading:

Healthcare Professionals

  • Diabetes Medicines

Patients and Caregivers

  • Diabetes Drugs

Medical Devices

  • CDRH Consumer Information for Medical Devices
  • CDRH Medical Device Safety and Recalls

Very glad to see these posts as I have noted the egregious device errors myself over the decades (from ALL manufacturers although Medtronic seems to be the worst). It is also true that the device companies are Marketing companies, first and foremost, they are profit-oriented, they can buy congress and they can buy the FDA. They are very powerful. My child and I stopped using these devices many years ago but, sadly, it seems even MDs believe the FDA and the marketing messages from corporate entities. Responsible, ethical, device companies are soon bought out by larger corporations and all human contact and empathy is lost. Thank you again for the posts.


I think its interesting that such a topic made it into Consumer Reports.
It seems too mainstream of a publication to publish on this very specific topic.

I found the article very interesting. I did not realize the extent of deaths and injuries by diabetes devices, and that is only the ones that were actually filed with the FDA. So many instances aren’t.


But even they make mistakes…

“To use one, you prick your finger with a small needle, squeeze a drop of blood onto a glucose test strip, and then insert the test strip into the glucose meter for its reading.”

Of course, this was true for early 1980s meters that required HUGE drop of blood and detected color change.


I contacted CR, both email and chat. Assuming they will correct…

1 Like

Ah, nostalgia. This is the season for it, I suppose. :wink:

Of course, the primary reason I often recall those very early glucose meters is I remember what it was like before any glucose meters existed. I did a terrible job of managing my blood glucose by essentially guessing. I truly suck at that.

They still haven’t corrected the mistake.

Like most organizations - the chat said they would pass it on - it will take a while to filter through to the system, get approved, assigned, and then done, particularly in this time of year.

Hehehehe, you all are funny. Made me laugh. :stuck_out_tongue_closed_eyes:

When you add complexity you add failure points. With MDI you can mis-dose yourself, but as far as the system itself goes there’s not a whole lot that can go wrong. There just aren’t that many components: Vial + syringe, or pre-filled injector + tip, and that’s it. The only failure point is really the human brain—operator error.

With a basic, standard pump, just off the top of my head:

  • Pump hardware: battery fail, mechanical failure (e.g., “frozen button” syndrome)
  • Infusion set and reservoir: inset rip-outs (can be unnoticed), bent canula on insertion, bad insertion location, 3-day limit, non-delivery due to kinked tubing or other cause.
  • CGM: bad or erratic sensor, blue tooth drop outs, sensor filament stuck under the skin, filament impinges on muscle sheath (ouch!), transmitter battery failure
  • Operator error: opportunities multiplied here as well. Numerous decision points that can be gotten wrong, or parameters shift due to metabolic changes: basal rate settings, bolus types that can be mis-applied (normal, square wave, dual wave), forgetting to restart after “suspend,” etc etc.

Hybrid system paradox: this was really brought home to me when I tried using the 670G. The ideal they’re working toward is something that acts like a functional pancreas, i.e., a black-box device that functions in a way you don’t have to think about, just like a real pancreas. But in practice, at least in Medt’s initial version, that entails locking parameter controls away from the user (black box!), while adding new potential failure points because the monitoring, feedback systems and insulins just aren’t really quite there yet (particularly the insulins). This can result in the worst of both worlds: it’s not working as well as the non-hybrid, the various fail-safes keep kicking in in ways that force you to have to jump to it’s tune rather than the other way around, and you can’t change things on your own to better adapt to your individual profile. The next iteration—Tandem’s Control IQ and the forthcoming 780G—seems to be evolving toward handing back more of the controls to the patient so you can dial it in to your particular metabolism. That seems a much better compromise given the current state of tech. You still have to know a lot about the principles behind everything in order to get the best performance, but once you do that it can perform more like a set-it-forget-it system.

Other causes:
cat/kitten playing with tubing and chomping it into many pieces !!
Worst case if overnight.


I agree with this sensibility about potential failure points and would add that (for me), the principal reason mdi works so well is because of a cgm (I mean, sure I could use strips 20x/day, but realistically, that isn’t as likely as looking at my phone every half hour), so the cgm potential failures can apply to the mdi model as well.

The article is interesting and generically suggests a lot of ways the manufacturers and regulators could make things better, more reliable, etc. They can’t control your cats (that’s funny @mm1!) but they can almost always do better.

Yup, in lots of ways CGM is way more important than pump technology—lots of people say they’d choose CGM over pump if it came to either/or. And this is why I don’t blame folks for sticking with MDI: just the complexity of a pumping system can be very off-putting, and you have to weigh that against the advantages

1 Like

I am a 670G/CGM user and have found ways to unlock parameter controls. For example, I use “fill cannula” to initiate a correction bolus when the pump’s algorithm prevents me to do so. However, by doing so, I have compromised other system processes such as correctly reporting of insulin on board and daily history of total insulin and Bolus insulin used.


Yeah, the user groups were full of work-arounds like that to trick the system into doing what it was, y’know, touted as doing for you automatically.