• Guest - w'd love to know what you think about the forum! Take the 2025 Survey »

The Fallacy of Average: How Using HbA1c Alone to Assess Glycemic Control Can Be Misleading

Interesting stuff. So, when all of us get our CGMs we can then get an accurate measurement of levels, it may take a while...
 
The following is from a Together 2 Goal/Diatribe presentation from June 2017.

Take a look at slide number 33 which describes the 'fallacy of average' nicely:

“If I stand with one foot in a bucket of ice water, and one foot in a bucket of boiling water, on average I am comfortable”

Slide number 37 looks at how important time-in-range is for all with DM:

“A Big Impact” on Daily Life, Rank Order: Many differences, but time-in-range is #1 for all

And slide number 41:

“The Many Faces of a 7% A1c – One Traditional Metric Cannot Tell the Full Story Anymore”

Which shows how 3 HbA1c results of 7.0% (53), tell 3 very different stories.

http://together2goal.org/assets/PDF/20170615.pdf
 
This has been an ongoing discussion in Diabetes Research for a good few years, and is well known in the "Expert Patient" realm.

Amongst T1s, the Libre has provided an insightful mechanism for identifying a lot of this. The real issue is access to data to allow Time in Range (TIR) to be used effectively. The Libre is one mechanism, but as we know, the requirement is that these devices become ubiquitous across all users to allow TIR to be a valid measurement, and the reticence of the payers to fund them is the biggest issue with access.
 
But Hba1c is a great tool when we track changes in it. Someone with type2 going from 80 to 50 without an incrase in drugs have clearly made great progress.

But we need to stop getting hang up on the different between someone with 48 and 52! As a Type2 I would much rather be told my fasting insulin, as I can track my post meal BG myself. hba1c is a tool not a god, so should not be workshipped and I never understood people who wished to predict there next result when they have access to much more useful and detailed data of their own.
 
Personally, it wouldn't bother me in the slightest if I never had another HbA1c test again. My HbA1c scores are nowhere near my own findings from frequent and systematic finger pricking and also from wearing Libre sensors. I have done a lot of reading on the subject, and can say this HbA1c reliability criticism is nothing new.
 
An ongoing discussion with my consultant that he should be looking at the avg BG and TIR, TNIR data from the Libre ahead of the clinic HBA1C measure. I daren't mention standard deviation yet!

Unfortunately, I assume reflecting the lack of access, the NICE guidelines only focus on the HBA1C as the golden measure for all.
 
Unfortunately, I assume reflecting the lack of access, the NICE guidelines only focus on the HBA1C as the golden measure for all.
It's more that the only evidence recognised by NICE is based on the DCCT study, which has very clear correlations between lower A1C and reduced relative risk of complications.

The science (and evidence) of improved TIR/reduced standard deviation and better outcomes is very much in its infancy (we've had "accurate" CGM for less than 5 years) and the cost of systems to monitor and prove this is relatively high. Until we have a better data set and cheaper monitoring solutions, as much as the cutting edge research states it frequently, these measures will not become the go to standard.
 
Also most people who get "improved TIR/reduced standard deviation" will also reduce their A1C and/or get the same AC1 using less insulin, hence when looking at groups of people A1C remains a very good predictor.

One large US study found that (in both Type1 and Type2) level of insulin usage was a better predictor of "length of life" than "great" AC1, e.g. people who used less insulin lived longer, however, AC1 is a very good predictor of complications (like going blind) that mostly don't kill people. Very bad AC1 is also a predictor of short life.

No one thinks a AC1 of 100 is OK, but is a AC1 of 45 any better than 50 on the individual level?
 
Also most people who get "improved TIR/reduced standard deviation" will also reduce their A1C and/or get the same AC1 using less insulin, hence when looking at groups of people A1C remains a very good predictor.

One large US study found that (in both Type1 and Type2) level of insulin usage was a better predictor of "length of life" than "great" AC1, e.g. people who used less insulin lived longer, however, AC1 is a very good predictor of complications (like going blind) that mostly don't kill people. Very bad AC1 is also a predictor of short life.

No one thinks a AC1 of 100 is OK, but is a AC1 of 45 any better than 50 on the individual level?

Hi @ringi do you have a link to the study you mention above? I am sure lots of people will be interested to read the details.
 
Just the title of this thread made me smile!! ;o)

And it is super refreshing to see one is not alone in this world to challenge the idiocy of chasing the HbA1c number as the holy grail. For years we have been able to pull out the bell curve with e.g. 2 std deviations from old meters but most physicians/nurses still appear shellshocked when having to explain what that means when looking at it. If they are the professionals, no wonder many diabetics remain stigmatized, relying on the few daily bg meter spot checks and the HbA1c reading they get every 3 months.
 
Thanks @ringi

The trouble with quoting the ACCORD study, is that the study used a cocktail of drugs to intensively lower blood glucose levels (rather than diet and exercise). Then, when the study participants had an increased rate of heart problems, the lower blood glucose levels were blamed.

However, one of the drugs (Avandia) used in the study was subsequently banned in several countries for causing heart problems.

Unfortunately, many people know about the heart problems, but not about the drug causing them.

There is more information on this here:
http://www.phlaunt.com/diabetes/35169265.php
 
If I recall correctly Metformin is the only drug the come out well on the Accord Study. (I think SGLT2 inhibitors were not included as they were not in common use at the time.)

Some people are now saying that part of the issue is that people got low HbA1c with high fasting BG, hence must have had very low BG in the daytime etc, maybe with lots of Hypos. What is clear is that HbA1c is not a predictor of the effativness of drugs for Type2.

Yet GPs get paid more if they get our HbA1c lower regardless of the harmfullness of the drugs being used........
 
I found this, which frankly disappointed me as, based on my HbA1c, I thought I was doing ok, now I am not so sure.
http://www.gpnotebook.co.uk/simplepage.cfm?ID=x20070614192625295600&linkID=70250&cook=yes

There are many such articles. I have a whole collection of them. It is very well known that red cell life span is crucial. If it is not around the average of the general population, then the HbA1c can be skewed. Then we have red blood cell abnormalities, such as anaemia, high or low haematocrit levels, hypothyroidism, and similar.

Trust your finger prick measurements over and above your HbA1c (if you test regularly enough that is)
 
There are many such articles. I have a whole collection of them. It is very well known that red cell life span is crucial. If it is not around the average of the general population, then the HbA1c can be skewed. Then we have red blood cell abnormalities, such as anaemia, high or low haematocrit levels, hypothyroidism, and similar.

Trust your finger prick measurements over and above your HbA1c (if you test regularly enough that is)

My blood tests have never indicated any problems that would affect HbA1c (at least any of the issues you mentioned) but I feel like my HbA1c is always way better than my fingerpricks/CGM would suggest. Obviously I will still try my best to get good finger pricks/CGM readings and a good HbA1c, but what can I trust? I don't feel like meters are always accurate (often finger pricks taken within seconds of each other give totally different readings), different meters give different numbers, and the CGM is only as accurate as the calibrating meter. Sometimes the trend on the CGM is more useful than the meter - a slight increase in on my meter may be due to inaccuracy, while the CGM has the overall trend correct and is validated by the meter on my next test. I've mentioned my concerns to my doctors but they reassure me that my control must be excellent because my HbA1c is. I feel like I'm being a bit paranoid but I keep worrying that I actually get bad numbers and there must be something wrong with me that makes HbA1c not work.
 
Trust your finger prick measurements over and above your HbA1c (if you test regularly enough that is)
That's the problem. From extensive testing when I was diagnosed 2 years ago, I now only test occasionally if I have some new food to test or out of curiosity. The HbA1c test may have its limitations but it is probably better than trying to get an average from a few blood tests taken say just pre and post prandial. Also, even if you have a non standard life of red blood cells this is presumably constant, so at least HbA1c trends would be useful.
 
I feel like I'm being a bit paranoid but I keep worrying that I actually get bad numbers and there must be something wrong with me that makes HbA1c not work.

I'm paranoid at times because my HbA1c tests are always higher than expected from copious finger prick readings and my Libre sensors, and I wonder if there is something wrong with my red blood cells. My current HbA1c is 44, which equates to an average of 7.3mmol/l over the same period. Even my very ordinary brain can see this appears impossible when my time out of range (ie above 7) is 1%. For 99% of the time I am between 4 and 7, mostly 5s. Even testing half hourly for 3 hours after a Sunday roast (my most carb heavy meal of the week) very rarely takes me into the 7s, and then only for up to an hour at most. I spend the whole of the night and up to evening meal in the 4s and 5s. So where does this average of 7.3 come from?

This is why I started researching matters, and why I take my HbA1c results with a pinch of salt. @Mr_Pot I do agree they can show trends.
 
Back
Top