For every BG monitoring system I add, I seem to also add more finger sticks… I attached a Libre today… then doubled my finger sticks. My family thinks I’m nuts, and I really might be.
I also have minimal body fat. I’d like to say “very little” but that really might be just to flatter myself. However, I’m certainly trim enough and need to insert my sensor into the abdomen area to avoid problems with muscle. I have a terrible time with my sensors, but I have a different idea as to why. First of all, based on some of the reading I’ve done on sensor technology, it seems that sensor accuracy is user-specific. So some of us just get the short end of the stick while others are camped out in “accurate”. Unfortunately, we might kind of be locked into our own historical performance.
I also have seen both sides of accuracy since I’ve started logging my numbers. I can get really intense with logging, especially high on coffee, and I track closely. I’ve seen my numbers be within 20% of my BG as much as 90% of the time (with the bulk being within 10%). Then I’ve seen it be within that 20% only 60% of the time (with only a small percentage falling within 10%). I’ve seen this with the same sensor in, same hydration, same everything… EXCEPT for stability. When my blood sugar is bouncy, my “accuracy” takes a hit. If I weren’t tracking the way I do and only comparing to the occasional BG finger stick, it would appear very inaccurate, but it’s really just lagging the entire time. And with bouncy blood sugar, sometimes my BG and SG cross paths, which would give the appearance of better accuracy, but it’s not… really…
I keep getting more monitors hoping one will be reliable, but it just seems the technology against the way my disease behaves, is not there yet.
Here’s my suggestion on that: reframe the way you think about what the CGM is. When we’re accustomed to using BG meters, we think of a CGM as an instrument to measure BG. But that isn’t what it is, and we’ll be disappointed if that’s what we require of it. What it actually is: an instrument and algorithm that, based on the history of measurements of interstitial glucose, show’s you it’s best estimate of the current BG measurement that would be consistent with that interstitial glucose history. The algorithm is compensating for the unknown and variable lag between BG and interstitial measurements as best it can. In a flatline scenario or a steady BG ramp the estimate should be quite good. During times of rapid change in trend, or disturbances such as “pressure lows” or dehydration, the estimates can be quite bad. Ok, so how can the CGM be a useful tool despite “quite bad estimates” in some circumstances. That’s where the reframing comes in.
Don’t think of the CGM as a BG measurement instrument. Think of the CGM as a tool to help figure out what to do to keep the BG in a good range. Let’s say I eat my usual breakfast with my usual prebolus. I expect to see my bg rise pretty quickly through the 130s, and level out and then turn down somewhere in the 150s or 160s. If I see it march through 130s, 140s, and into the 150s with no hint of leveling off, then I’ll take more insulin because for some unknown reason I need more. And it doesn’t matter if my actual BG is lower or higher by 30 mg/dL because I just watched it march upwards 40 or 50 points and it’s not showing any sign of turning back down, so in any case I need more insulin. If the CGM says I’m steady at 87 and I’m actually steady at 123, that’s an enormous percentage error, but does it truly matter? I say it won’t hurt me if I think of it in terms of “What do I need to do now?” To me, this is the foundation of the sugar surfing approach: look at the changing shape of the BG graph, and based on how it differs from your expectations, add insulin or carbs to guide the graph in a good direction. That’s where the CGM becomes an exquisitely helpful tool rather than a mediocre BG instrument.
I absolutely agree with you 100%… Across the board, I think we are understanding this thing as the tool it is. Which brings me to the question… why did I say that when I really do know better? It’s MOSTLY because I do a lot of things for no good reason at all, but I also think that when it comes to understanding this technology and even this disease, I’ve taken the slow, scenic route. In fact, for 12 maybe?? Of the almost 15 years I’ve lived with this, I tested RARELY. I also owned a sensor but didn’t wear it often. Turns out sensors appear really accurate when you wear them once a month and confirm by BG once a week. Then even when I did start testing, it took a while for me to understand what was going on… It just felt inaccurate, and that could be a little frustrating… It wasn’t until wrestling with auto mode though that I really started tracking these numbers, and that was another level of disappointing. In fact, I think it was all the difference… when my sensors of old were “off”, it was a nuisance. But with this one actually delivering insulin… or withholding it… it became more critical that it not have such a delay. Yes… that’s it. It was needing this pump to be able to make appropriate treatment decisions for me that made me start to want so badly a sensor that had no delay. I know it’s not how they work. It’s why I can’t be in auto mode.
As far as sugar surfing, I have never read it, but it’s definitely how I run the show. I have learned how to read my sensor values so that they are helpful in letting me know what’s upcoming. I understand the difference between a 83 and 91 could reflect a rebound that’s taking place. If I have enough insulin on board, I’ll just get up and start to move. If I don’t, I’ll test and treat.
Yes… this is the difference. When you’re using that number as a general guide, it doesn’t. But when it results in your missing out on another 25 minutes of insulin, it does.
And now that I’ve defensively tried to show all I know about sensors, I’ll go sit and think it over… the bit about changing my expectations. I really do know a lot. It doesn’t change the fact I want it to be an up to the minute measure of BG. Excellent point.
Based on your reports and others similar, I wouldn’t even attempt 670G auto mode for myself. Because I would get into a serious discussion with that very expensive machine, and by serious discussion I mean something involving a large, heavy hammer and a determined expression of displeasure.
Let’s hope that a much better algorithm comes out. LOOPing has given me lots of reasons to be optimistic about the possibility for a system that surfs as well as we can do manually, but that, unlike me, always pays attention and never delays taking action because I’m up to my elbows in soapy water or some other completely valid excuse that leads me to decide I’ll just deal with my BG later.
I’ve used Dexcom since 2011 or 2012. I have learned for me:
My low back is the most accurate, followed closely by my upper arm…and a distant third is my thigh and an embarrassing fourth place goes to my abdomen. The first two spots I listed are most accurate and most timely. My stomach just can’t get it together. It’s a mess. Totally unreliable for anything D-related.
I ignore the arrow during fast changes. It takes three data points for it to change angles (i.e. fast drop vs slow drop vs level, etc)…so I look at the 6 hour horizontal graph when things are “dynamic”. The 6 hour view gives me the best idea of whether this is an OMG moment or not. 3 hour view is too subtle…12 hour view makes everything look OMG.
Just my two cents…
I know you’re talking about a different CGMS, and I’d have to try your theory against my own CGM to see if it even makes any sense, but whether or not it does is not critical to this statement: I love that you have a system and that you appear to know it like the back of your hand. I’m going to go look for the OMG effect right now just because it sounds like fun and cross my fingers it’ll work for me, too.
