This is a follow-up on my review of the Dexcom Stelo continuous glucose monitor (CGM), one of the first marketed to consumers. This CGM is available without a prescription and designed to be used by people with and without diabetes-like conditions.

I have now used 5 sensors, 15 days each, and put together the data.

There is inter-sensor variability, which is a challenge

The image at the top of this post shows that from sensor to sensor, there’s a different baseline reading, even in the absence of dietary changes. This should not be a problem if a CGM is not being used to determine what blood glucose “is” versus how it changes based on diet and/or exercise. For reference, I am on a (mostly) low-carbohydrate diet and exercise regularly. I do not have diabetes or prediabetes. I do have the “Lean Mass Hyperresponder” (LMHR) phenotype.

Why is this a challenge? Because a person without diabetes has a very narrow range of “normal” glucose readings. In my case, I set my expected range from 70 mg/dl to 110 mg/dl. There happens to be an “emerging” body of data on what this range should be. This article helps a little in that regard.

These two images illustrate the potential problem. These devices are engineered to measure a wide range of glucose values, from the lowest of the low, to the highest of the high. Out of the box, then, the glucose range of 70 – 110 mg/dl appears very tight using the default axes:

Day to day, though, this is not what a person without diabetes sees. They see something more like this, when the axis is right sized to the selected range:

This pattern can be confusing at best, demoralizing at worst, to a person without diabetes. They may believe these are wide swings, when they are not from your body’s perspective. Also, if someone is using a CGM to understand what the baseline is, the inter-sensor variability makes this determination difficult. What we might arrive at is a place similar to one we did with pedometers (remember those?), where we learned that measuring something (steps) did not result in better health.

Implications for the future

There’s no question in my mind that these are impressive devices that give us important insights into our health. At the same time, it’s not clear that these insights may improve our health. In addition, all of this assumes the sensors work as intended, which is not the case in looking at online discussions of their utility. There are many stories of failed sensors, improper placement, and other technical issues.

For my part, I was able to learn what I wanted to from this run, and will use these devices only sporadically on an as needed basis. It’s great that they’re available on demand without needing a prescription.

As always, comments and questions welcome.