Dangerous flaw in Medtronic 670G

Just made an account specifically to share this because it is dangerous and I’m not confident that Medtronic will fix in a timely manner:

I had the perfect set of unfortunate circumstances today that made me realize the 670 does NOT alert at all when insulin flow is blocked- except when a bolus is attempted.

I had a hectic day at work during which I ate very little and only low/no carb foods, so I never tried to give a bolus until the afternoon when my BG was up to 346 and I started feeling bad (which was unusual because of how little I had eaten and how much activity I was doing all day). I tried to correct with the pump and only then did a no insulin flow alert show up. I couldn’t immediately change the set so I quickly gave a correction with an emergency syringe to tide me over. From that correction, my BG came down, I soon changed my set and everything seemed fine. I might not have connected the dots except that the SAME thing happened AGAIN.

Coincidentally, I had a second faulty set insertion, didn’t eat or try to give a bolus for another 3 hours… then I felt sick, checked my number, tried to correct…and received the “insulin flow blocked” message again. I immediately changed out the set, this one has been fine…but the events of the day opened my eyes to the very likely possibility that the 670 just isn’t programmed to alert if basal insulin is blocked. I called Medtronic to tell them about it (their representative was lovely and very understanding and patient!) However, based on all of the many problems and bad reviews this pump has had, it might be a while before this bug is fixed (if ever). I just wanted to share so people know this is a possibility and can be mindful of it!

1 Like

ALL of their pumps take forever and a day to alert to a blockage thru basal activity only; you’ve got to do at least a few units bolus for the sensor to react to a blockage. I’ve been using MM pumps since 1996, 5 models in all, and they are all that way. There is no “fix” for this.

Think about how much pressure it takes to push the plunger in the reservoir. 2 o-rings that are tightly fitted and of course the small orifice thru which the insulin flows. If there was very little normal backpressure, the sensor could more easily distinguish when there is a blockage, in a more timely fashion. But that’s not the case.

1 Like

That will always be the case. Because the amount of insulin put in by the basal (or the microboluses if you are using Auto Mode) is so small, the pressure change is not enough to trigger the alert. It’s like when you weigh a bag of jelly beans at the candy store, adding 1 bean will most likely not change the displayed weight, but adding a whole scoop will. Do you use CGM at all? If you do (assuming you have it correctly calibrated), you should be able to receive alerts when your BG gets that high (actually, I would probably suggest setting the alert lower than 346, but whatever number you choose). The CGM also lets you look to see your number (the sensor glucose values are slightly different than the BG values, but they are similar enough that you will be able to see if you are, or are starting to get, high) at any time if you feel like you may be high or low. If your doctor did not get you CGM (for the 670G it would be the Guardian 3), I would definitely ask them about it, because it can be a very useful and important part of managing your Diabetes.

Surely they could put in a software trigger that says “Hey check your infusion site” in a more formal way.

Sorry for your high blood glucose troubles. I have a few points to make and only make them in the interest of better decisions next time.

I agree that a pump with a low basal rate will take a long time to create enough back pressure to sound an alarm. I don’t think this is a design problem with the 670G; I think it’s design challenge for every pump on the market.

This makes it all the more important to monitor blood glucose regularly, especially when on a pump. Since we pump users don’t take any long acting basal insulin we are vulnerable to quickly sliding into ketoacidosis. It’s one of the main reasons I wear a continuous glucose monitor. I want to know if my glucose is trending high.

Whenever my glucose (fingerstick or CGM) is above 250 mg/dL and especially when it crosses 300 mg/dL, I only deliver correction insulin with a syringe. That way I can be sure that the insulin was indeed delivered. These high levels will almost always cause me to change my infusion site, too.

I have been put into some vey unhealthy and unsafe situations due to my diabetes with both very low and very high BGs. Diabetes is nasty and doesn’t play fairly. We all need to treat it with the respect of the deadly foe that it is.

I apologize if this seems preachy. I’ve used insulin pumps since 1987. None of them have been perfect but in the end they are only tools that we choose to use. Granted, the manufacturer must deliver us a safely designed and built device. But we must also assume some responsibility. We each are the single biggest factor in whether our outcomes with pump therapy will be good or not. The pump plays an important role but our brains and experience are vital to our success.

I suspect that you’ve already drawn some important lessons from this experience. Good luck going forward!

1 Like

Hi Terry, thanks for your reply. Your post made me realize that I should make a few clarifications and share some info that I have learned in the day since this happened- since the point of my post was just to warn others using this particular model. (I know you are a pump user, but I suspect you might not use this specific one- apologizes if I assume incorrectly).
As another very experienced pump user who is careful and in control- I did initially feel frustrated (though I’m sure no offense was meant) with your pointers about respect and responsibility, because I think it’s missing the point.

So, I realized that it might be important to note some more specific details- my first set failure was not a newly inserted site, it just failed after working perfectly in the day and morning before the trouble started… as you know occasionally can happen. I had no reason to be extra concerned about it as I would with a new set, and I never went an unusual or negligent amount of time without checking during the whole experience. Regardless, when new basal insulin is not being delivered, especially with a fair amount of physical activity, as I was doing yesterday- the difference can be unnoticeable until several hours have passed with no insulin on board. I suspect it was very possible that the insulin flow was blocked even during my last BG check before the one that discovered the problem (it just was not high enough to warrant a correction yet). There is truly no way to know, even for a vigilant person.

I did not have a failure of responsibility yesterday or make poor decisions (and therefore did not learn any “important lessons”). Rather, I had an experience with a pump model that did not live up to very basic and reasonable expectations. I only feel confident saying this because i have plenty to compare it to- I have happily used four Medtronic pumps before this one and have received much more reliable and efficient alerts in the past.

Based on new information gathered today, it seems the problem is how high the cumulative pressure must be before it alarms. For many people, their basal insulin needs will take many hours to reach this limit. That could be an issue with other pumps too- I have no idea- but it was not an issue with prior Medtronic pump models, in my experience.

People who have switched from a previous Medtronic model to this newest one should not expect a timely alarm as in the past (which is exactly why it’s dangerous… it is reasonable to expect the same or improved precision and safety in the newest technology, but sadly, it looks like the opposite is the case here). People who require basal rates on the lower end should be especially aware of this problem/flaw/situation/whatever we want to call it!

On a related note, I am not currently using a CGM, but not for lack of trying! Unfortunately, after many sensor failures and hours wasted on the help line to no avail, I’ve taken a break from trying to use Medtronic’s CGM. I know I’m not alone in that, so that’s another safety net that many of us trapped in this dissappointing new system do not reliably have available.

Anyway, I’ve already said too much…Thanks just for offering community during this exasperating and worrisome moment in time.

1 Like

I’m sorry that I made an incorrect assumption about your experience with insulin pumps. You have more experience with insulin pumps than I thought. I do not have any experience with the 670 so my prior comments may be out of line.

You made a post to warn people about a possible flaw with the 670G. I should have accepted your report at face value and not widened it to include other issues.

Like you, my experience with Medtronic pumps is long. In fact I’ve been using a MM722 for the last few years. While I have had some false no delivery alarms with Animas Ping models, I have not had any “no-delivery” trouble with the Med-T models .

My analysis that there may not be anything wrong with your pump’s no delivery alarm may have completely missed the mark.

I still think, in a general sense, that when using a pump, it’s good practice to correct high BGs with a syringe. Others may disagree with this and do differently, I understand that we each have the autonomy to make our own decisions.

Sorry I strayed into offering unsolicited advice. And my lack of experience with the 670G should have restrained my comment. I hope your pump performs better for you or that Medtronic acknowledges the problem and fixes it.

1 Like

I like that statement, it shows that no matter how good our pumps and CGM may be (or, in some people’s opinions, not be), we still need to include our own senses and instinct when taking care of ourselves.

As you say, others may disagree, and I am one of those people. My reason for this is because of how injections can affect the data used by Auto Mode on the 670G. It also sometimes confuses me because, even though it may be the same amount of insulin that the pump would be putting in as a correction, the fact that the basal (or Auto Mode) is missing sometimes makes it hard for me to factor everything in until the Active Insulin expires. But this philosophy obviously only applies if you are planning on inserting a new infusion set reasonably soon (in my mind, less than an hour). If I received the message while somewhere which I did not have quick access to either one (for example, the movies), I would still wait to change the infusion set, because it would require the same amount of time to get back home for either one. But as you say, others may have different philosophies for this kind of thing, so please don’t take this as saying you are doing things wrong or as an insult if you do them differently, this is just my reasoning behind the way I do it.

Like @Terry4, I will do correction by syringe when very high. Particularly when I think a bad infusion site may have contributed to high BG. But I then disconnect the pump, and do same bolus so that the pump knows about it. And yes, I just wasted a few units of insulin. But pump IOB will show it. I am using older Medtronic 523 pump, so not worried about auto mode, which may be different in how it handles this.


That is clever. A parallel pump “air-shot” keeps the pump informed. I’ve heard of this tactic before but have since forgotten. Thanks for reminding me!

1 Like