I've got a CGM - what kinds of values are normal?
If you have a CGM, what kind of ranges might you expect to see? Since CGM’s are so new, we don’t know for sure, but we have some data to draw on.
These studies have been done in people with normoglycaemia, and each show the baseline A1c of the people included in the study. This is helpful because right now, the most reliable evidence we have to draw on to understand the link between glycaemia and health is A1c because we have long-term data in tons of people. Conversely, there is only a limited amount of short-term data on CGM-derived glycaemic measures in normoglycaemia. And we have zero data on CGM-derived glycaemic measures in populations without diabetes and any long-term health outcome.
But we can do a bit of educated guessing and try to draw inferences from all this data. For example, we can probably say that people with an A1c under 5.7% are just fine from a CVD perspective in terms of their glycaemia. So if we measure glucose using a CGM in people with an A1c <5.7% under their usual diet/lifestyle conditions we can probably get a pretty good indication of the degree of glycaemic excursions that occur in people whose glycaemia (again, based on A1c, which has very very strong evidence behind it) is probably just fine.
In other words, we can see what kind of glycaemic excursions are compatible with a (probably) worry-free HbA1c.
Here’s the data:
In this study, they recruited 53 participants (age 7 to 80 years) all of whom had an HbA1c of <5.7%. The CGM they used was a Dexcom G6, which was blinded so the participants could not see their own glucose readings. They found that a median of 2% (IQR: 0.9 to 3.9) of the glucose sensor readings were above 7.8mmol/L (140mg/dl), but in people aged over 60 years this went up to a median of 4.1, IQR 1.3–8.6. Now the CGM would have taken a blood glucose measure every 5 minutes for 24 hours - so 288 times a day. 2% of 288 is ~6. 1% is ~3, and 4% is ~12. So in this group of people aged 7 to 80 years, with normal glucose tolerance, a median of 6 glucose readings using a CGM were over 7.8mmol/L, but in some people this went up to 12 readings. In people over 60 the number of readings over 7.8mmol/L basically doubles.
In this study, 80 people without diabetes and with no history of diabetes, a fasting glucose level ≤5.4 mmol/l (97 mg/dl) and HbA1c <6.5% were recruited. A Medtronic Minimed was used as the CGM. A quarter of the participants had glucose which was above 7.8mmol/L for at least 75 minutes per day. Three participants had a glucose value of more than 7.8mmol/L for 5 h or more per day. Seven participants had a glucose level that reached 11.1 mmol/l.
In this study in 24 people without prediabetes or type 2 diabetes and with a mean HbA1c of 5.0 ± 0.2%. The CGM was a SCGM1 by Roche Diagnostics. The maximum sensor glucose seen in this study was 171 mg/dl (9.4mmol/L) after a standardised meal which had 50g of carbohydrate, and 168mg/dL (9.3mmol/L) for free choice meals. In this study people spent 99% of the time with sensor glucose under 7.8mmol/L (140 mg/dl).
In this study, 74 participants with A1C ≤6.0%; fasting blood glucose 70 to 99 mg/dl; 2-h oral glucose tolerance test (OGTT) level ≤140 mg/dl. The study found that sensor glucose values were >140 mg/dl for only 0.4% of the day, respectively. This study had an interesting design which was to allow the use of 3 different types of glucose monitor and for different periods: (a Guardian Clinical by Medtronic for 3 days OR a FreeStyle Navigator for 5 days, OR a DexCom SEVEN for 7 days. Then weirdly, they thought the Dexcom have readings that were unusually low and high, so they discarded the Dexcom data…….(plus they said the DexCom data sensor malfunctioned too much. In any case, this type of study shows how many factors can affect glucose readings with sensors - and why we REALLY need to interpret them with caution.
In this study in 32 participants with normal glucose tolerance (Mean A1c: 5.3 +/- 0.3) found that seven participants had glucose levels above 11.1 mmol/l (200 mg/dl) and that participants spent on average 42 min per day at glucose levels >7.8 mmol/l (140 mg/dl).
Caveats
Note that when people are studied they tend to - subconsciously or not - change their behaviour. If you’ve ever been asked to complete a food diary for a study, you might remember suddenly being very aware of what/how much you were eating, and decided against that cake. You might also want to “be good” so you get good results on your glucose monitor/blood pressure monitor/whatever outcome the nice investigators are studying. So it would not surprise me at all if people in these observational CGM studies ate better diets during the study than they would normally do, such as eating half a pizza instead of a whole one. So the sensor glucose readings in these studies could easily be (obviously I am speculating and I don’t know for sure) an underestimation of typical glucose peaks.
Summary
In the absence of any evidence to the contrary its reasonable to suggest that post-meal excursions into the 9-11mmol/L range occurs quite a lot in people with a perfectly fine A1c, and is probably a normal part of human physiology and nothing to worry about.
If we do get data that shows the contrary, of course I will reassess and update here!
More reading:
And if you’re thinking “it’s not just the peaks, it’s the peaks and troughs!” - you need to look at glycaemic variability!!!” then click here.
If you’re thinking - “I want to be cautious and keep my glucose as low as possible anyway” - read here.