A1c and CGM not matching

just got back from my doc- A1c is 9.1 (214 avg over 3months) but my CGM says over the last 28 days My avg (it checks every 5 mins) is 166 big differance!

Some might have you believe that there’s some easy lookup table that uniquely matches A1C with average bg but really there’s a lot of scatter in the relationship.

Just for reference, look at these scatter plots from the DCCT comparing A1C with average bg, and you’ll see that an A1C of 9.1 with an average bg of 166 ( = 9.3 mmol/liter) is not an outlier:

I dont know what I am looking at does this go off of a thumb print? lol I get the A1c side but whats the 0-50 side?

the zero to 50 is the european canadian measurement method mmol/l vs US mg/dl

Converter here

Tim, Thanks for the data - where did you find these charts, I would love to revirew the source data - I never knew that there ws so much scatter in the data and do you know the difference between the Scatter Plots a,b,c vs d

oh ok got it - thankx
but what about the thumb print vs. the line whats that?

It is not a thumb print. It is a scatter plot where thousands of data points are plotted. The line is the computer best fit regression line of the data, the center of the scatter so to speak. so you could read the plot that for a given BG value, the most likely A1c is the point on the line that intersects the given value and to see the range of the likely variation one would look up and down to the boundries of the thumb print

wow - ok - thank you Sir. you seem very well informed thankx for inparting knowledge - very helpful
thankx again

I believe those scatter plots were done using the BG at the time of appointment - so it is not surprising that they would be so variable for T1 diabetics whose BG can vary from 30 to 300 within a few hours. Even if done as an average, it would have been from only a few tests per day and not necessarily representative.

A better predictor and closer to the question being asked was done more recently using AG (average glucose) from CGM and finger-stick data compared to A1c. 166 still looks to be within the range for A1c of 9.1 (although it is on the edges of the range).


Your CGM is giving you a more realistic average.

The A1c (contrary to popular belief) is NOT giving you an average. The past week is weighted most heavily (I have seen a chart several times that says the past week can account for up to 17% of your A1c value). The ‘weight’ of each week decreases through the past 3 months. (This is because red blood cells don’t necessarily all survive to 90 days, so some of the ‘old’ ones are dead and gone…thus they can’t count in your A1c, even though they were glycated once. On the other hand, your newly glycated red blood cells from the past week should all be around still.) So having a super 2 weeks, 3 months ago is going to make very little difference if the past 2 weeks you’ve been averaging 200. (Just an example)

The A1c is also not highly controlled. Labs are allowed to be off by a certain amount and still have a ‘valid’ result. I believe it’s .7% (so your A1c could realistically actually be between 8.4 and 9.8) (someone correct me if I’m interpreting this wrong). And finally, you may be a low or high glycator (the rate at which you glycate blood), which could change your value to make it unexpectedly high or low.

your post Kari confirms my thoughts that i have been telling my fam and friends for the last 5 yrs- glad that I am not alone in this thinking -

I dont freak to much - my mom is more the one who worries -(yes at 26 and married with 2 kids , my mom still is the main worrier) I know that i workout/ run 4+ times a week and eat pretty good and sleep 7-8hrs and all that other stuff - my blood pressure is 118 over 64 with a resting heart rate between 43-52 - i am healthy, happy, and doing the best i can-- its just that point of i have never in 12 yrs had an A1c below 8.4 that gets me- but thankx for your input Gangst’a it helps to hear your thoughts

Thanks for the update and the more accurate plot - yet this plot isw still only a gestimate as the date was not complete for teh 90 days the study startes that the average number of days for cgm was only 13 and for fingersticks was 39. I would really love to see a study of 90 day cgm averages vs A1c - I wonder if we could get Tudiabetes to start such a study. Remember also that the apparant variation also includes the calibration error whish is based on our own fingersticks, so I think that the true BG variationfor a given A1c is probably less than the graph indicates.

thats funny you say that about the docs fitting more in - my doc this time looked at my CGM print-off and ignored the A1c (kinda lets me know my doc might be good) good ol’ Mid-west small town Bible belt action- my doc is one of the best i have had for the last 7-8yrs