We did a sensor change last night and the last 18 hours of data have looked like output from a random number generator. Last night Dex said 39 – he was 141 according to two, clean-hand finger sticks. This morning Dex said 186 – he was 76 according to two finger sticks. We typically bolus off Dex but obviously right now that’s impossible, and I’m a little scared to have openAPS running while it’s so off. i know the first 12-24 hours of data can be really wonky but this is beyond the level we typically see.
So, can a bad sensor reform, so to speak? Will this just be a bad sensor for its remaining life, or do you think it could get highly accurate in the next two or three days?
As an aside, we’ve had a run of pretty crummy sensors lately – anyone else experiencing that?
This just happened to me within the first 12ish hours on a new sensor—I had a LOW CGM reading when meter said 98-99, and a 298 with a double arrow up when the meter said something like 141 twice in a row. I’ve never had one that off before. I almost just yanked it then, but it seems to have come in line though, a couple of days later, so I’d give it a chance.
Yes, we have indeed had some wacky readings initially, only to see it come back. It’s very, extremely rare though. Have you considered a restart? I think I would try that. I don’t give up on sensors too easily, mostly bc of the unpleasantness of insertion, although I do not enjoy two hour black out periods either!
I ran into something pretty close to what you’re describing. It was in the first 24-36 hours of a new sensor, which is always supposed to be a little flaky, but this was way out of whack, with the sensor failing to register lows in particular–I rely on my Dex to monitor for lows while exercising and that’s when it became evident to me that something was seriously wonky. But it did settle down after that and I kept using it for the usual two weeks or so. When I pulled the sensor I discovered a significant amount of dried blood around the insertion point and on the underside of the sticky patch. I’d evidently hit a capillary. I remember thinking at the time that it was interesting to know you could have a bad insertion like that but the sensor would eventually settle down and you’d still get the full use out of it.
I have seen numbers get better…but your issue is intriguing. A few months back I, too, had a run of bad sensors. Everything is back to normal now, but at the time it was extremely frustrating. That is my biggest fear of the CGM directed artificial pancreas system. After several calls to Dexcom, new transmitter, new sensors, new insertion sites–my wife and I came to the conclusion that it was a bad batch of sensors.
Question: Is this sensor from a new batch/lot?
Upon reflection, our Dexcom delivery was left outside in direct sunlight and the box was noticeably hot to the touch (although Dexcom assures that wouldn’t be an issue).
My advise? Rule out the easy stuff first, and if it persists call and get a replacement sensor from a different lot # than what you currently have.
I have this problem throughout all my Dexcom usage. It doesn’t matter where I’m wearing it, when in the session it is, it’ll happen. In a one day reading the difference between my meter and either the Dexcom or the CGM on my T:Slim is off up to 95%+/- half the time. It’s really infuriates me when it wakes me up telling me I’m either 250 or 65 and I go check and discover I’m 100. Or during the day it goes off and says I’m 42 and I’m really 115. Or another time it’ll say I’m 175 and I’m really 50. Thankfully I have my body signals that I know and trust. If my endo didn’t think the CGM was so great I’d forget about wearing it.
Like the others who’ve commented so far, I’ve had some wacky results the first day or two with sensors that do ultimately “settle in.” But the one time I had numbers that were as WAY off as the ones you describe, I bailed and got a replacement from Dexcom (and was pleased they were happy to provide that).
I wonder sometimes if the sensors just have a really short shelf life and are prone to instability the closer you get to the expiration date. I haven’t kept records to test that, versus the possibility that some batches just come out with higher probability of poor performance, versus the natural uncertainties that result from things like bleeding at a capillary like @DrBB mentioned, versus the multitude of other insertion site variables, etc…
At the risk of jinxing ourselves, my daughter’s Dexcom sensors have, with the exception of a very few over the past 2-1/2+ years been scary accurate, some of them right from the start. We’ve never, knock wood, had one that was wonky and never came into line after 10 +/- hours.
I’ve had the same thing happen and the sensor has come into line within 24-48 hours. I was traveling and had already called Dexcom for a replacement sipped to my hotel. Bonus sensor!
Like @DrBB, I’ve noticed a high degree of correlation between wacky readings and a blood -fouled site. If crazy readings continue into the second day, I cut my losses, pull the sensor, and ask Dexcom for a replacement.
Part of the problem with Dex readings that are far from fingerstick numbers is the system then has to reconcile a reading like “Low” which is under 39 mg/dL and a fingerstick of 120 mg/dl. When you insert that calibration number, the Dex system will often come back with another calibration request or just question marks (???). Some sensors can recover from this, some can’t. I’m not patient enough to waste my time coaxing an extended session out of a wayward sensor. It can be done, sometimes. But I don’t like that game. I cut my losses and move on.
And then you have people like me, who are the complete opposite. If a sensor is inaccurate I’ll calibrate three times 15 minutes apart (I got this strategy from Dexcom support) to get things sort of back on track. (If this doesn’t get things back on track and it’s during the first week, then I’ll call Dexcom for a replacemetn, but I’m not sure it’s ever happened during the first week.) I also find that calibrating three or four times during an inaccurate day can help the sensor get back on track (but I only calibrate twice if accurate). If it’s still inaccurate and/or giving a lot of ???, then I will restart a sensor. If the restart goes bad as well, I’ll restart again. If the sensor still doesn’t come back then I may pull it, but I may give it one last restart. I’ve had a sensor that gave a “Sensor Failed” alert and then constant ???s after two restarts only to have it come back for an entire week of accurate readings after the third restart, so now I often can’t give up on a sensor without giving it that one last chance. This is how I get by using about 1/3rd the number of sensors each year as Dexcom recommends. If my insurance covered sensors (and, knocks wood, there may be some potential there that is being investigated) I’m not sure I’d spend quite as much time trying to beat a dead sensor back to life.
I’ve done this, too. I feel like I’m arguing with a dumb machine at times, but I have had some success. I will generally be patient for the first 24-48 hours. After that, I get tired of the dance.
I’m glad that the Dexcom has been so helpful to you. I think it’s a life-saver.
Yes, I didn’t want to add that to my comment without knowing if it was relevant but doing a calibration when mine was that far out of whack was a concern. Could be problematic for the accuracy of the rest of that session at least. I assume the algorithm resets when you initiate a new session, though I don’t really know if that’s the case.
Interesting–the general rule seems to be that the fewer calibrations the better, but given the source this seems worth considering. I assume you’d only try this if it were consistently far off and other factors like dropping or rising, or being very high or low weren’t involved.
I second DrBB’s experience – the time or two I’ve had this happen, I’d hit a small vessel. If this happens, the sensor won’t work right, because it’s not bathed in interstitial fluid, but rather blood.
Sorry to say, you need to discard it and use another.
thanks everyone for the advice! We hung in there and by about 24 hours, it was pretty spot on! Good to know something so abysmally bad can recover. Next time we may call Dex for a replacement if it keeps being wonky, just in case.
I’ve done this too at Dex’s recommendation. Works sometimes if the sensor isn’t on its last legs. See also discussion at this thread:
dexcom-trial-reading-about-30-different-from-meter/55876/8
My sensor that had the extreme out of whack numbers ended up still losing signal much more than typical, despite being mostly accurate after the first 24 hours, so I yanked it when the first week was up, instead of using for 2-3 like usual. Sure enough, some dried blood. First time I’ve seen that, and first time I had such out of range numbers as well as as much missed signal time.
This is how I get by using about 1/3rd the number of sensors each year as Dexcom recommends. If my insurance covered sensors (and, knocks wood, there may be some potential there that is being investigated) I’m not sure I’d spend quite as much time trying to beat a dead sensor back to life.