“Non-diabetic” glucose levels

As I’ve written before, my own glycaemia has been described by some doctors as “approaching normoglycaemia”, and my last lab HbA1c result came in at 33 mmol/mol (5.2%). I’m not going to pretend that I’m “normal” (and I’m not the example in this article) but personally I have been feeling that my BG profile has at least been moving reasonably close towards it.

Recently my partner wore a FreeStyle Libre sensor. She doesn’t have diabetes, but some of her relations do have Type 2 diabetes. We had a Libre sensor that was otherwise going to expire before it was used, so this was an opportunity to check on her current glucose profile. She also got to experience part of my world: wearing a medical device continuously!

Libre data

We used a Libre Reader to initialise the sensor and to gather the data. We had a premature sensor removal on the second day, but got to use a replacement sensor. So at the end we had almost 16 days of data.

Afterwards we downloaded the data and analysed it both with Abbott’s FreeStyle Libre software and via Glooko’s Diasend to get an overview. And yes, she did give me permission to share this data with the wider world.

My own experience in the past with a Libre showed me that the values reported by the Libre Reader are an approximation that doesn’t always match a BG meter or a calibrated CGM, but it’s better than nothing. Even though the instantaneous values don’t always match a fingerprick, I’ve found that the overall data averages from a Libre Reader do provide useful data.

Day by day

Over the last few years my partner has had to endure me showing her my CGM data (on my phone or watch) as things happened during our lives. Whether it was my BG going up or down after various meals/exercise, or me marvelling at how it didn’t change in some circumstances. Now she got to do the same to me!

But on the whole we simply got to marvel at how “boring and flat” her data was! 4.9 and 5 were numbers we saw a lot of.

We do try to eat reasonably healthily in our house, although as my partner doesn’t have either T1D or coeliac disease she does have opportunities to indulge in a few treats that I tend to stay away from. There were a few occasions where her BG “spiked”, and they were all attributable to specific foods. Some sample days:

It’s impressive what having a functional pancreas can do for your blood glucose control!

The overview

Blending all those days together we get an AGP plot:

My path lab’s HbA1c reference range for someone without diabetes is 25-38 mmol/mol (4.4-5.6 %). So I don’t think there’s much chance that her next lab test will indicate diabetes (or even “pre-diabetes”)!

The 3.9-7.8 mmol/L range used on that graph was somewhat arbitrary, but is the goal range that I have been using for a while for myself.

I’m happier with that than some older data of mine:

Diasend’s analysis

The AGP plot looked similar, but the statistics present some interesting numbers.

It’s noticeable here that the Libre Reader summarises the data into 15-minute samples.

The average shows up as 5.3 mmol/L. I’m prepared to put the difference from the 5.2 reported by the Libre software down to rounding. It still converts to an estimated HbA1c of 30 mmol/mol (4.9%).

The standard deviation of 0.8 results in a Coefficient of Variation of 15% (0.8/5.3).

The TIR percentage for the 3.9-7.8 range was 97.8%. For 3.9-10 it was 97.9%. But most of the times the data dipped below 3.9 mmol/L were attributable to compression lows.

Of course, this was only a fortnight’s data, and my partner hasn’t volunteered to wear a sensor on an ongoing basis. But I’d like to thank her for helping me understand what “normal” might look like!