I tried to get into the 180 day trial but was too late at Rainier in Seattle area, it has already started. There might be spots open elsewhere, see the link that lists other labs:
https://clinicaltrials.gov/ct2/show/NCT03808376
I had a phone conversation with the lab and the 180 day study sounds pretty intense. In hindsight I’m not sure I would have participated if there was room…once monthly there is an all day session at the lab with an IV hookup and frequent blood draws (if I remember correctly). Although they pay you to participate and there is a free CGM for 6 months so that would be a great way to try the new CGM for those who have the time.
There was a 2018 trial that assessed accuracy within the first 30 days (which is my main issue now on Day 4). According to Senseonics, “The PRECISION trial was designed to provide additional data on 1) the accuracy of the Senseonics CGM System within the first 30 days after Sensor insertion…” and “The PRECISION study MARD is slightly higher than PRECISE II due to the additional days 7 and 14 accuracy testing, longer duration accuracy challenge on days 1 and more low glycemic challenges. Across the board in both studies, the ‘Percent Within Readings’ achieved or exceeded the 85% 15/15% level considered extremely accurate.”
I am hoping to find the time next week to search out a copy of that 2018 trial and see how my results compare. I have read several reviews by users who had similar wonky numbers in the first week or so.
Do all new devices have to undergo testing in the US before FDA approval? I would think that a clinical trial should be pretty much the same anywhere in the world. On the other hand, maybe there something unique to the US that’s throwing off the results? Deep fried mars bars, anyone?
Seriously though, is there something I’m overlooking here?
Very quickly it gets legal. It is not difficult to imagine a number of devices which are very much in the grey area. Many are obviously under the jurisdiction of the FDA.
As you are coming with a European point of view, it may be not quite as natural to consider the US FDA point of view. Safe to say, “NO”, the US FDA is not going to approve something just because it was approved over in Europe.
The following link gives a good idea of the types of devices which would fall under the purview of the US FDA:
It depends what the trial needs to prove. In the US, the trial needs to prove that the medicine or medical device is both safe and effective. Perhaps in some other places the standard is just freedom from harm.
Most countries in Europe and many others around the world use CE mark to determine if the device can be allowed in the country. Most countries have some additional screening requirements, but CE is required and central to approval. To gain CE you need to prove that your device is safe. However in recent years CE is moving more towards the FDA and requiring some efficacy data. In the US you need to prove that you are safe and effective. This is a much higher hurdle and requires additional studies. Many companies start in Europe with CE while they are collecting the data necessary for the effectiveness study to gain approval in the US and access the big $.
The Eversense seems to be settling in starting about Day 3/4. I’ve been getting decent agreement between the three BG measurements although Eversense numbers are still a little jumpy. Here is a comparison of Day 5 before breakfast data !!
Note that the Eversense graph shows some scatter that is not realistic (that’s what I call jumpy) and the G5 smooths out that scatter. Other reviewers have noted that the algorithms are apparently different. I hope to discuss with technical support next week also. The scatter seems to trigger false alerts for the “predicted low glucose” which I have now turned off.
So far things are looking good for the big test planned for Sunday: hockey with iPhone packed away and no G5 receiver strapped around my belly. I will be counting on the alerts on my arm to avoid going low during hockey or high immediately afterward.
I’m now on Day 8, the insertion site seems to be completely healed over. I’m getting some variable accuracy, seemingly not tracking high BG very well but doing fine once I am solidly into target BG range.
I ended up going high after my hockey game yesterday afternoon, brought the BG down, bolussed and overate for dinner and had a long sleepy high BG overnight. The screenshots are from Day 8 AM and show how slow to respond to my BG dropping down from that high BG. The Eversense was quite slow to respond to BG coming down to within range. Once I got back in range and calibrated, it seemed quite accurate for the rest of the morning. (note the G5 was off too before calibrating).
The only issue I can think of is every one of my Eversense calibrations so far have been within the BG range of 75-150, I have never calibrated during a high BG. Not sure what to do about it but will see how things go for the next few days.
Still trying to decide if Eversense is reliably accurate for me…my G5 transmitter dies tomorrow so I will be limited to comparing Eversense to fingersticks unless I start a new transmitter (which I had not planned to do). I’m thinking 10-20% inaccuracy when high is not the end of the world, as long as it does not tell me 193 when I am at 134 (44% error). Might be a calibration issue as noted.
I am hoping to get past the accuracy/calibration questions soon and post some info about the Eversense app. So far I like it a lot although it takes some getting used to. Will keep everybody posted.
Yeah, us diabetics aren’t a trusting bunch, it will be nice to see if you come to trust it, and honestly, a bit of error when high isn’t the end of the world as you mention. It would be nice if they published the linearity of their glucose response, that would help visualize some of this.
Day 9 started out with Eversense missing a low…Time to decide if I should start a new G5 transmitter (I am on day 111). The high readings in late night and waking are concerning. I have requested technical support to do some diagnostics on the sensor.
I share your concern! The low was caused by a correction bolus for the false high BG alerts at 2:45 AM. I was a little sleepy/lazy and did not check the Dexcom or fingerstick until after I had bolussed. My mistake, I did not heed the warnings to check with a fingerstick before dosing from a CGM. I had grown accustomed to ignoring that warning with G5 but sometimes we learn things the hard way.
I’m also wondering if maybe I was dehydrating overnight both times I had the false high BG readings. Both nights were after hockey games and although I felt pretty well hydrated when I went to bed it’s possible that I was not. Maybe the Eversense is a lot more prone to reading high when we are dry?
The official word from Eversense is that sensor data sometimes does not stabilize until 10 days after insertion…I am not there yet but getting close.
Maybe if I had more time I would try to figure out Spike but to this point I have been OK with stretching G5 transmitters to 112 days. Plus I need a reliable CGM to work for me Weds. night (hockey) so I will probably raid the single G5 transmitter on my meager stockpile and continue using the Dexcom as a cross check. I’m going to need a CGM backup in early July anyway when I get the second Eversense 90 day sensor inserted, to get through the initial week to 10 days of use. My original plan was to use a Libre as my backup but Dexcom will provide a better backup scenario for the rest of the trial.
I’ve started a new G5 Sensor and transmitter so I’m covered. I’ve had a few calls with Eversense technical support about my data accuracy and have found them easy to get on the phone, knowledgeable and informative. So far a positive experience working with them on this. They looked in depth at my data on the cloud site and offered suggestions.
First thing I’ve learned as they reviewed my data was they encourage more calibrations “as needed” beyond the two daily required calibrations when my data is not close enough to fingersticks. This is contrary to advice I’ve learned about Dexcom for the G5 but they confirmed it a few times. The idea is to swing it back into a better calibrated condition to account for whatever is affecting it’s readings. There does not seem to be a hard and fast rule for these extra calibrations but they did confirm that the algorithm uses the previous four calibrations.
There also is an easy way to log my fingersticks in the app so they will show up on my data plots but do not affect the calibration (unless I enter them as calibrations). I will be trying both of these new habits for about a week and see how things go.
Does that mean it uses exactly four? So when you take a fifth calibration, the oldest drops off and now using the most current four? Which I would assume.
If only using four (a small number IMHO) data points for calibrations, it would be worth asking if there is value to be had in trying for a range of calibration values. Rather than the last four calibrations being in the BG range of 80~120, would it be better if you grab a calibration at moments of opportunity to calibrate (if reasonably possible) at lower BG, mid-range and higher BG?
