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

How to guesstimate HbA1c

Eldorado

Well-Known Member
Messages
168
Location
Buckinghamshire
Type of diabetes
Type 1
Treatment type
Insulin
Dislikes
Diabetes.
Apologies if this question has already been answered, but does anyone have an idea how to guesstimate HbA1c results? If my average BG is 6, I'm wondering what my next HbA1c might be. I'm trying really hard to get mine down from just over 8%(too high I know) and have been using the Aviva Expert. Great meter btw. If any meter can be great, that is! Thanks.
 
I have failed to guesstimate my HbA1c on each of the 4 tests I've had since I started testing (Type 2 diet only)
I test a lot, including tests before and at an hour, 90 minutes, 2 hours and 2.5 hours after meals when necessary, plus fasting and bedtime. My own averages are always lower than my HbA1c by about 0.7mmol/l.
 
Run a rolling 90 day average in Excel or an App without a bias to taking blood sugar levels. You need a good cross section of tests through the day.

I predict mine within 0.1% of my actual Hba1c result. Using Aviva Nano. I tried a TrueYou blood test meter and the readings were on average 10-15% lower than the Aviva.
 
Since all our home test meters produce snapshots, there is no way that an average of readings will accurately reflect an average of blood glucose across each 24 hour period. (It is why I want a Freestyle Libre)

Nor do I think that the HbA1c is particularly accurate. Several things affect it, including anaemia, diet variations across the (approx) 3 month period and whether your own body replaces blood cells quicker or slower than expected. The results also are weighted towards recent bg readings.

I think all these tests and readings are wonderful, invaluable tools, but I see them as useful to show trends, not absolutes.
I don't agree with using an HbA1c as a diagnostic tool for diabetics. But it is a good guide of how controlled your diabetes is.
 

Good point! Individual test meters may well screw the results too.
 
Brunneria said:
Since all our home test meters produce snapshots, there is no way that an average of readings will accurately reflect an average of blood glucose across each 24 hour period. (It is why I want a Freestyle Libre)
Click to expand...
But bear in mind that the sensor inaccuracy in the Libre (or any other CGM) also skews Hba1C predictions. My current sensor is currently running consistently 1.5mmol/l high, which means that for the two weeks of data it will load an incorrectly weighted value on to the Hba1C calculation..
 

Yes. The reason I want the Libre is more to do with tracking my DP (when I'm asleep), and catching the height and timing of the high point after food.

I've been watching the Libre thread with interest, and have no illusions about the thing.
 
So my last Hba1c at the hospital was 5.8%. My forecast was 5.93% with an average BS of 6.877mmol over 90 days. Divide 6.877 by 1.583 then add on 1.583 as a formula. This is based on 354 tests.

Some meters would be more accurate than others from what I found out using the TrueYou Mini which tells me that my blood sugar is about 15% lower than it actually is.

There may be variants in blood cell life cycles and other factors that show deviations. But if Hba1c is the industry standard then i'd adjust for it if possible depending on what the error rate of blood tests taken at home is.

I have over 8500 blood test results on my computer and statistically disected it.

I've closely forecast my Hba1c in the last 6 years since analysing on Excel.

I've been type 1 for over 26 years and computerising my results from an analysis background has improved my control. But that works for me, everyone needs to find the right tool i guess and a lot of self exucation.
 
For what it's worth, out of idle curiosity I've found that using the Diabetes UK converter, my 3 monthly BG averages have been a pretty close prediction for my last couple of Hb1Ac results. But since I'd be lucky to get anything other than an annual check from our Practice diabetes doctor (my GP doesn't think this is frequent enough and gave me my last test), I prefer to rely on my meter to provide me with my most useful information... I believe my Hb1Ac test doesn't give actually me any more accurate information over time than my meter and Glucofacts software do, and an annual Hb1AC tells me b****r all. But watching for trends and any highs/lows will tell me more about managing my diabetes than just an occasional average anyway - so as far as I'm concerned now, my Hb1ac result is just for looking pretty on my diabetes medical records.

Robbity
 
On an individual level then the HbA1c is affected by all those things mentioned earlier. Nevertheless it's a good risk indicator and in type one at least the evidence suggests that it doesn't much matter how it's achieved People with fairly stable levels over a day in the DCCT had similar risks to those whose blood glucose went up and down more during the day ; just as long as they had similar HbA1cs
Re working it out, when I used sidiary to record my levels, it used a similar calculation to the one used by Pseudo bob. It was pretty accurate for me though obviously depends on the number and distribution of your tests through the day.
There is a meter on the market that is supposed to estimate HbA1c ( I saw it advertised in the pharmacie the other week and thought it was new but a quick check shows me it's been around in the UK for over a year http://www.diabetes.co.uk/news/2014...se-monitor-that-estimates-hba1c-90860366.html
 
Last edited by a moderator:
Also you can work out a distribution analysis, look at the interquartile range plus frequency of excursions below 4 mmols and above 10 mmols and when they occur in the day. There will always be some variability, it's just a matter of capturing it and recording it.

90 day averages are good for monitoring longer term risk, but run the averages over 4 weeks so you can see a more tangible response to changes in blood sugar patterns. Trend the peaks and nadirs and highlight your average daily range.

I think anything from 4 to 7 tests spread through the day should enable most people to leverage a good degree of control and give a pretty clear picture.

All i can say is that you have to be actively on it. Waiting for the doctor to guide you every 6-12 months doesnt really help. Thats why i think all the android apps are great to help so many diabetics gain visibility of this day to day condition.
 
I'm not too good at maths, but I can and do use all the features on my meter to set (my own) targets, and check weekly, monthly, 3 monthly averages. It will also tell me if I'm off target and for how many times over any of these period.

Robbity
 
Hi Robbity,

It's what ever really works best for you individually.

After working in accounting, finance and analysis I decided to apply that to my blood sugar results and from there educated myself more about statistics. When I first put a spreadsheet and graphs together six years ago i had to figure out what the hell it was all telling me. Anyone can put an average together, the hard work is understanding it all. Constant incremental adjustments everyday.

Targets are a great way to manage, you adjust you improve then adjust some more. Then something in your life changes and wham blood sugars are back out of whack. Constant monitoring will allow you to correct pretty quickly.

Elevated Hba1c seems to be the risk indicator of acute diabetic complications and mortality rates. The longer i've had diabetes the more i've learnt that this is a statistics game of numbers.
 
My software does precisely what @PseudoBob77 says and then applies the formulae against the 90 day average. You can also hovver over the points in the graph for the 7 and 30 day trends and it will give you an estimate. For me I see little difference between 7, 30 and 90
 
Hi Andrew,
In your case if you can arrive at a reliable estimate then thats a good benchmark.
 
So if you know what your average variation is you can adjust for your results if necessary if you know its going to be a certain percentage difference.

Good find on the diabetes journals Indy51
 
You must log in or register to reply here.
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn More.…