Here’s a graphic for you to use as a quick-reference guide to put HbA1c numbers into context, and show the relationships between HbA1c, Blood Glucose, and GMI.
Unlike many of the “A1c chart” graphics that have been doing the rounds over the years, I’ve tried to strike a compromise between too-little and too-much information. I’ve also made it relevant to all types of diabetes, not just T1D.
Regarding the selection of HbA1c levels:
- 4.4% and 5.6% feature as the edges of the “reference range” for people without diabetes.
- 6.5% is a level that can be a trigger for T2D diagnoses.
- 7.0% is the level that clinical guidelines advise to not exceed for long, as the risks of health complications have been shown to rise significantly above that.
eAG (estimated Average Glucose) is a value often shown on laboratory pathology reports.
It is the inverse of the eHbA1c calculation which estimates HbA1c from average BG.
In recent years many of the CGM vendors have been changing their software from using the traditional eHbA1c formula to the newer GMI. Sometimes still labelling it as “HbA1c”. I have previously written about GMI and its issues. I’ve tried to keep eHbA1c and GMI separate here, as they are different things.
Probably neither formula is perfect for everyone, but personally eHbA1c has always been a better match than GMI for my own lab results (ever since I started collecting detailed CGM records in 2017).
I hope this is useful to you!
I would caution against the under 7 is fine compared to under 6 argument. Historically when the only treatments where older insulin and Su drugs for diabetics it was difficult to maintain a glucose under 6 percent without going low. There are studies that show an a1c of under 6 can for example reduce heart disease. In addition there are questionable studies showing no benefit under 7, for example an over 400 pounds patient on multiple meds such as avandia where included. An a1c 6.9 for example maybe be ok for 5-10 years as a new diabetic of for certain months where it was difficult to get control but not say 30 years.
I’m not keen on the “under 7% is good enough” line either.
However I do think that multiple levels of target are fine, and 7% has been shown to be so much better than 8% for example.
So as an initial goal <7% works. But I do think there's value in going a little lower if we can.
Personally I prefer my levels to be in the green band on this graph, but a decade ago that seemed unimaginable.
This chart matches my experience well. Out of these three metrics, I’m a fan of simply using the mean BG if a CGM is available.
A1C surely has been useful for research purposes, since it was available even without frequent fingerpricking or a CGM. However, in some studies where they urged participants to fingerprick frequently, it has been shown that the fingerprick BG predicts complications with better precision than A1C [1]. So I assume CGM mean would also be superior to A1C.
I think it makes sense to estimate average glucose from reported A1C values in studies. But conversion other way around makes less sense to me.
And of course, TIR could be even better metric but it’s a separate topic.
[1] https://link.springer.com/article/10.1007/s00125-007-0883-x