Standard Deviation – What is it and Why is it Important to a Diabetic?

Standard deviation (SD) is a bit complex, but in simple terms, it's a statistic that tells you how tightly all the various examples are clustered around the mean in a set of data. When the examples are pretty tightly bunched together, the standard deviation is small. When the examples are spread apart, that shows you have a relatively large standard deviation.

For a diabetic, knowing your SD tells you how well controlled your blood sugar is.  For example if you have an A1c of 5.0% - many would look at that number and think “that’s awesome!”.  But is it?
How you obtain your A1c is just as important, perhaps more important, than the number itself.  If achieving an A1c of 5.0% means that you have very large blood sugar swings, say from 40’s to 300 (large SD), a lot of damage is being done to your body due to the rapid swings. An A1c of 5.0% is really not very good in this scenario.  However, if you achieve your 5.0% with blood sugar levels ranging from 60’s to 120 (small SD) there is no damage due to the tight control and you have a healthy blood sugar average.

So what is a good SD?

A very good SD is 25-20. A great SD is under 20.  Where did these numbers come from?  A non-diabetic with a healthy A1c will have a natural standard deviation of about 5-15 mg/dL.  That's how tightly our pancreases control our blood sugar.  I tested this personally... my Non-D daughter wore a CGM for 2 weeks and her blood sugar average was 79 mg/dL with an SD of 6.  I wore a CGM for 2 weeks and my average blood sugar was 91 mg/dL with an SD of 11.  Trying to achieve that tight of an SD with T1 diabetes just really isn't feasible... at least not consistently.  The <20 - 25 SD is a bit arbitrary, but it is realistic for someone that wants really great control. The lower the SD the better because the tighter the control you have.  Remember the whole point of knowing your SD when it comes to diabetes is to avoid blood sugar swings.  Minimal swings means less damage to your body!

How do you know what your SD is?

If don’t have a CGM, you’ll have to calculate it manually.  There’s an app for that!  Just search "standard deviation" in your app store and you'll find free ones.  It’s really simple… plug in all of the #’s from your finger pricks (FP) for a day and it will give you your SD.  Do that each day and then you can average your SD for the week or month. You can also use an Excel formula.

Here are some examples in the Google Play store:

If the app you use lists both Population SD and Sample SD, for blood sugar purposes, use the Population SD.  It looks at all the BG values you entered. The Sample SD is only looking at some (a sample) of the values you entered.  An SD calculator looks something like this:

After you calculate, you'll get your SD results:

The downside to FP's is that your set of data is very limited.  There are 1440 minutes in a day.  If you FP 12x a day, there are 1428 minutes left that you don’t know what your blood sugar is doing.  But on the plus side, the data you do have is very accurate!

Another way to get your SD is if you have a CGM.  A CGM sensor is reading your interstitial fluid (ISF) every 5 minutes and is estimating what your blood sugar is.  It sends this data to your CGM receiver which can be downloaded to pull reports that calculates your SD for you.  The pro is that you have lots of data.  The con is that CGMs can be off; sometimes a lot.  They often don’t read the lowest lows or the highest highs.

I have a lot of T1 friends that love their CGM reports, but a downside is that the data can be skewed because it’s all interstitial data.  Their report may show their A1c is 5.0% with an SD of 15, but then their lab A1c is 5.5%  The reason is because the CGM is off, and if the A1c is off, so is the SD.  The point is to be healthy, not simply achieve a number.  I don’t care if a report shows an A1c of 4.5% with an SD of 10 if that’s not REALLY what is happening in ones body.

So what do you do?

My opinion is that if you have a CGM you combine the best of both worlds – lots of data points (CGM) plus accuracy (FP's). Get your CGM SD from your report.  Then take the SD from actual FP's and average the two to get a truer SD.  If you use Dexcom, you’ll have to combine and average this manually.
Medtronic CGMs also record FP's so when you pull a Carelink report it will show you your FP SD and your CGM SD… all you have to do is average them out.

Summary: looking at A1c or SD alone can be very misleading, but together you have a powerful tool.  I hope this helps simplify SD and helps you to understand why it's so very, very important to your diabetes control and health.

There is another formula that I’ll cover in another post that adds a 3rd layer and provides even more insight to overall diabetes control… %CV (Percentage Coefficient of Variation).  Sounds funs, right?