I hope this helps.

Some people need to recall that all measurements have or should have an error included therein (often not stated but implied) that is particular to the tools of measurement and the user. This is something in addition to watching one’s significant figures, and using them properly.

For instance, if I use a cheap wood ruler to measure the length of an inert metal bar, and it looks like 13 and 2/16 inches, I probably cannot make the claim the length is 13.125 inches in length with 5 significant figure. It might be that the accuracy of the ruler is plus of minus 1/8 of an inch. So I then might say that I measured 13.125 inches +/- 0.125 inches (as the error). Further, perhaps two sig figs is best, so the metal bar is then merely 13 inches +/- 0.12 inches error. But I cannot presume that the tool (the wood ruler) or my use of it is accurate and precise. I might by my method introduce some small error too in precision, which is different from accuracy.

When you use a home glucose monitor, it has an error that is inferred with every measurement, unique to the meter. Some meters are off by as much as a few digits up or down. If you want to demonstrate this take ten measurement in a row at approx the same time using only one meter. Here’s a sample data set one might get: 132, 135, 127, 130, 125, 133, 130, 131, 138, 128 mg/dL. You cannot say your true value was any one particular figure but it is safe to use the median or the mode as 130 mg/dL. However, a better way to say it might be 130 mg/dL (2 sig fig) or 130 mg/dL (3 sig fig) +/- error. The error in this case involves the sum of the sq’s which might be in the end +/- 3.7. So maybe your true number was 126, or maybe your true number was 134. It’s all pretty much the same number given the imperfect tool you’re using to get a value.

Now we’re using two different tool, a home meter and a CGM to get to different value to compare, both with limitations, and both not perfect, and both with errors implied therein.

So 124 reading on one device and 132 on another is not any surprise whatsoever, really. Plus the G6 might lag slightly behind the finger stick too. I am not advising you on which number to use although I have a method myself. If you think you have this down to a highly accurate and precise means of determining blood glucose level and injecting insulin, without error, one is kidding himself. For one thing, if I told you to show me some particular amount of insulin, let us say, 47 units, I guarantee you if you did it enough times some would be 46, some would be 47, some would be 48, etc. . . and that is more than a 2% variation just for this. Also site have different characteristics, some are better and some are worse than others. So, 47 units in one site might be different than 47 units in another site in terms of response. And then, do you wish to assume every vial or pen of insulin has identical activity – because they do not. Everything is in a range and has a certain error.

Basically, your two readings of different values are reasonable and within the error range. 134 and 124 might be considered the same value for practical purposes here. it’s almost like saying 129 +/- 5. It’s not the same as that but you get the point. That’s about a 4% error in this example. I do not know the error range on your meter. It is available information. There are different methods and practices to evaluate statistical significance of particular errors. I wouldn’t be surprised at all if a meter has a 3 or 4% error rate that they state in the literature. In fact, I would expect it.

This presentation only discusses one aspect of the CGM issues, not the whole story. BTW, I have a new G6 and haven’t used it yet. I hope to set it up today. I am not expecting perfection.

Do not let it drive you crazy.