Glucose variability -- an underrated metric

Type 1 and LADA
1

I’ve written about this topic before but I consider successful management of glucose variability provides incredible benefits. How do I measure glucose variability? I use a CGM and follow the Standard Deviation (SD) statistic as a reasonable proxy for glucose variability.

I’ve seen this referred to from time to time but it is not often emphasized in treatment protocols. Time in rage or time hypo along with A1c or average glucose are more often the subjects of discussion for glucose metrics. Glucose variability is often left out or is an after-thought.

A SD of 20 mg/dL or fewer is the target that I use. What is it about the quality of an SD of 20 or fewer? When SD is higher than 30 or so, at least for me, I find that it provokes more glucose variability. A low SD tends to produce more periods of low glucose variability. Like many things in life, success tends to breed success and failure tends to lead to more failure.

It’s ironic to me that focus on glucose variability using SD will often lead to more time in range, less time hypo, and lower glucose averages.

My animal-based low carb way of eating forms the basis of my ability to live with an SD of 20 or fewer. I’m aware of individuals who eat many more carbs per day than me, yet still manage glucose variability at low levels.

What is it about a maximum SD of 20 mg/dL that makes it desirable to me? This relatively low level of glucose up and down provides a calmness and quality of life that I find a huge benefit. I suspect it reduces cortisol levels and leaves me unperturbed and better able to handle minor levels of stress that life naturally contains.

Does anyone else monitor glucose variability via SD as a primary means of managing T1D?

1 Like
2

Of course, but you and I have discussed this for years.

When I’m implementing “rough” system controls, I’m primarily using this, alongside max and mins. The classic use case is backcountry skiing where I’m doing something highly anerobic over a full day. I run tests and work my way up to something that is “good enough.”

1 Like
3

Given the variability in CGM output I don’t feel that either SD or min/max is reliable. In fact I don’t use SD at all for a 24 hour period. I certainly use min/max, particularly for backcountry though I haven’t had a real opportunity to do that for 20 years. I use min/max with downhill, with any outside work and, when I did it, with scuba. Also, of course, operating a vehicle on a highway.

These are short term considerations; I didn’t interpret @Terry4’s post as referring to those, rather to the long term, say 30 or 90 day, performance.

The advantage of SD as opposed to TIR is very clear; CGMs are wonky, they give systematically bad readings, but SD is not affected by those. I’ll try to give a simple example:

Suppose my mean BG is 150mg/dL. This corresponds to an HbA1c of about 7%, fairly typical for well controlled T1s, indeed a target for endos according to the Google AI.

Now suppose I’m using a CGM, silly me. The CGM will read all over the place. Ok, take a “target” SD of 20mg/dL; this means that about 2/3 of the time my BG will be in the range 130-170mg/dL and that 95% of the time it will be within 110-190mg/dL. More than 99% of the time it will be within 110-190mg/dL. (Google “68–95–99.7 rule”, read the page corresponding to your religion.)

Now if I use TIR my results suddenly depend on an absolute, not an average. Rather than measuring the variation TIR considers <average+/-variation>; a shift to the average changes the TIR because my extremes are much less frequent than my close-but-no-cigars. Check out the graphs on whatever web pages you ended up from the google search in the last paragraph.

Ok, sorry, this is seriously arithmetic. I might jump out of my bath with no clothes on (boy, was that guy a dick!)

It’s also completely irrelevant to the life of any sane T1. The HbA1c gives us a long-term result which, so far, is irrefutably correct and that describes, so far irrefutably, our long term prospects. TIR is useful because it tells us how many times we got into danger-land; no one wants to be around us when we are high or low.

SD? It’s completely accurate, it’s a number which suggests other numbers. Taken with HbA1c to get the true average it may be useful as a diagnostic technique for all those endos on the planet who sat through the stats course and paid attention. You probably don’t want one of those as your endo.

We can use it. We can use TIR. It’s a lot of work; I’m sticking with HbA1c and a AID system that stops me going too low or too high.

2 Likes
4

I probably use it most as a measure of sensor accuracy.

5

I mostly use the 14-day report to look at glucose variability using SD. The 30 and 90 periods are similar. My current 14-day SD is 21 mg/dL. The 30-day is 20 mg/dL and the 90-day is 21 mg/dL. For me, the 14-day report represents the longer periods well.