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

Finding my HbA1c myself

lucylocket61

lucylocket61

Expert
Messages
6,394
Location
Wrexham
Type of diabetes
Type 2
Treatment type
Diet only
I have taken all my meter readings from the past 57 days, and found the average is 6.6

They range from 5.4 to 8.2(just one bad day) as daily averages. I think my HbA1c would now be 5.5.

Is that right? If not, how would I go about getting the right figure. I got this figure by using the HbA1c calculator here. (I think I did it correctly)
 
Your meter readings are just a snapshot and don't cover the whole day, so your estimation is probably not accurate as you don't know what your levels have been during the times you haven't tested. The HbA1c looks at the whole time period. I think to do your own HbA1c you'd have to buy one of the kits like Swimmer and Borofergie have.
 
I use a similar procedure to estimate my HBA1C but it's prone to error. The formula I use on my spreadsheet is the same as the calculator I think, but it's;

({30 day average} * 18.05 + 46.7 ) / 28.7

but there are a couple of problems with this. The first is that my readings have to be pretty complete. I don't test at night and I miss out a lot of lunchtime readings, cos I'm working. Without a very complete set of readings there may be spikes or lows that I'm not considering.

also, my new A1C meter's instructions suggest that the HBA1C reading is loaded so that recent levels have a disproportionately higher effect on the HBA1C result than levels that occurred at the beginning of the period. If this is true then a calculation based on 30 day averages (or 57 day averages in your case) would be less accurate.

For what it's worth, my 30 day average is currently 5.75 and this calculates out (using the formula above) as an HBA1C of 5.24 - my meter yesterday gave a result of 6.1 . All of these figures are probably inaccurate in some way - but as long as I apply the same calculations in a month's time and compare, I can still track my own progress.
 
swimmer2 said:
also, my new A1C meter's instructions suggest that the HBA1C reading is loaded so that recent levels have a disproportionately higher effect on the HBA1C result than levels that occurred at the beginning of the period. If this is true then a calculation based on 30 day averages (or 57 day averages in your case) would be less accurate.
Click to expand...

Yes it should be highly biased towards recent weeks. Some people suggest that it mostly depends on the last 2 weeks.

If I was doing it* then I'd use a 70:20:10 rule. Average = 0.7 * month 3 average + 0.3 * month 2 average + 0.1 * month 1 average

*I don't really measure my BG often enough to make it worthwhile.

Forget about all that. Great work Lucy!!!! :clap: :clap: :clap: :clap:
 
Post by borofergie » 2 minutes ago
Be very careful - don't mention HbA1c calculations in front of Pneu or Grazer, or they'll wop their rather extensive spreadsheet calaculations out.
Click to expand...

Thats fine. I have a super-techy teenager here who did my spreadsheet and would be able to cope with techy stuff.

I just feel so upset and desperate about how hard it seems to get the info I need from the NHS to monitor and control my Diabetes. And if i dont monitor my blood sugars, how will I know if I am doing right or wrong?

For me, its not good enough for the nurse to test my pee with a stick once a month and pronounce my blood levels as normal, then trust her judgement. Then wait for symptoms to appear to get more tests. As I understand it, waiting for symptoms means damamge is being done?

And how will I know if my medication is right or wrong? Being told that "Diabetes is a slow disease and nothing drastic will happen in 12 months between tests" is not reassuring me. And I dont have another surgery I can go to.

If I am over-reacting or misunderstanding something, I hope some on here will tell me.
 
lucylocket61 said:
For me, its not good enough for the nurse to test my pee with a stick once a month and pronounce my blood levels as normal, then trust her judgement. Then wait for symptoms to appear to get more tests. As I understand it, waiting for symptoms means damamge is being done?

And how will I know if my medication is right or wrong? Being told that "Diabetes is a slow disease and nothing drastic will happen in 12 months between tests" is not reassuring me. And I dont have another surgery I can go to.

If I am over-reacting or misunderstanding something, I hope some on here will tell me.
Click to expand...

You're not over-reacting and you've understood everything. You're just coming to terms with the fact that you've been given a bunch of terrible advice by people that you should be able to trust to look after your health.

It's no small thing to ignore advice from your HCP and instead follow that of a bunch of loons on the internet. You did it and it worked, but it still leaves a bad taste in your mouth realising what might have been.

Your BG readings are great Lucy, you should be very proud of yourself.
 
I think you have to be a bit of a Maverick to go against the NHS doctrine but thank heavens there are enough of us about to be able to spread an alternative scenario to the 'you are doomed' message that too many of our fellow Diabetics are being handed.

Up the Diabetes Liberation Front comrades. :D

_DIABETES_LI_10060.gif
 
Doesn't work. You need to know your TRUE average for the calculator to work, not just the average of your readings. For example, what happens to your BGs in beween readings? When is your peak reached (it's not after 2 hours) and how long does it last for? What happens to your BGs overnight? What is the average of your BGs under the curve from before you eat 'till after it's back to "normal"?
Lots of people use formulae to arrive at an approximation, but often any correlation to their true HbA1c relies on a bit of luck.
You need to work out your TRUE 24 hour average BG for the formula to work, which means you need either continuous BG monitoring or a mathematical model. Even then, the conversion to HbA1c is for the average person, and none of us are actually average. Even the formula itself is under some challenge with the American organisations using a different formula.
 
Grazer said:
Doesn't work. You need to know your TRUE average for the calculator to work, not just the average of your readings. ...Bleat...bleat...bleat.....Even the formula itself is under some challenge with the American organisations using a different formula.
Click to expand...

Don't say I didn't warn you!
 
borofergie said:
Grazer said:
Doesn't work. You need to know your TRUE average for the calculator to work, not just the average of your readings. ...Bleat...bleat...bleat.....Even the formula itself is under some challenge with the American organisations using a different formula.
Click to expand...

Don't say I didn't warn you!
Click to expand...

You've always been jealous of my sinusoidal graphs!
 
Grazer said:
borofergie said:
Grazer said:
Doesn't work. You need to know your TRUE average for the calculator to work, not just the average of your readings. ...Bleat...bleat...bleat.....Even the formula itself is under some challenge with the American organisations using a different formula.
Click to expand...

Don't say I didn't warn you!
Click to expand...

You've always been jealous of my sinusoidal graphs!
Click to expand...

If I knew what a sinusoidal graph was, I might be too. :lol:
 
I disagree with you Malcolm although it doesn't actually make any difference whether I do or don't as I end up agreeing you wont get an entirely accurate result.

HbA1c is based on what's called a regression analysis model. Without looking things up exactly I think they built the model something like the following. Take 1000 diabetics and get them all to test 7 times a day (one fasting then a before and + 2 hour after each meal). Over time each diabetic builds up a profile of results. Plot those results for all diabetics on a graph and work out the straight line that best fits the graph. Like all straight lines it will have a formula in this case as Swimmer says the relationship can be described by ({30 day average} * 18.05 + 46.7 ) / 28.7.

So if you test 7 times a day and are close to the average of those 1000 test diabetics then the relationship will give you a good fit between BG and HbA1c regardless of what your 24 hour average is. However the obvious flaw is that all diabetics will not necessarily be near that average and therefore the relationship wont be as good. I was testing 7 times a day from diagnosis to my last HbA1c test yet my HbA1c was 0.4% higher than I was predicting. All I've done now is keep the same formula but add in the 0.4 difference so that next time if may be closer.
 
I think that everyone is right, but that you're all also wrong.

I agree with xyzzy, that there is a measured correlation between HbA1c and average BG. You can also formulate a correlation between height and typical weight, but it doesn't mean that people are tall because they are heavy (correlationa and causation).

This correlation leads to the misperception that HbA1c is a measure of average BG, and it isn't really. It measures accumulation (glycation) of sugar molecules by your blood cells during spikes. HbA1c increases when your BG is high, it doesn't decrease when your BG is low, so it isn't a true average.

My HbA1c reading of 4.5% correlates to an average BG of 4.6mmol/l which is completely wrong (my average would be much closer to 5.5mmol/l). But I get this because I almost never spike (my BG doesn't really go above 6mmol/l even after eating), so my blood cells don't accumulate any glucose particles.

On the other hand, their are plenty of non-diabetics with HbA1cs of 4.5% who eat a regular diet, spike quickly after eating, before returning to a BG level of probably around 4.5mmol/l as predicted by the correlation.

The 4.5% to 4.6mmol/l correlation is based on their "typical" response to eating and not on my strange VLC diet.

In order to get a true calculation of HbA1c, you'd need to resolve all the spikes, and then do some complex maths. You'd also need to account for the fact that people glycate at very different rates, such that people with identical average BG scores can differ by as much as 1mmol/l in HbA1c.
 
xyzzy said:
disagree with you Malcolm although it doesn't actually make any difference whether I do or don't as I end up agreeing you wont get an entirely accurate result.

HbA1c is based on what's called a regression analysis model.
Click to expand...

Those (the majority) who are bored with technicalities should change channels now.

Ah, yes, you're quite right, although I know you meant to say "HbA1c conversion from average BG is based on..." because of course the actual calc of HbA1c is not based on regression but a complex bit of science deriving the ratio of glycated and non-glycated hexapeptides, hence why it was expressed as a %age before the IFCC standardised method. Anyway, as I say, you're right about regression, but it depends what model you want to use. The original DCCT model, which our website uses, gives an entirely different figure for HbA1c than the American association model which, I think, Phoenix suggested was more reliable.
DCCT was BG=(A1c*1.97)-4.28
ADA is BG=(1.583*A1c)-2.52

However, there was a further calculation called the Nathan formula. This was
BG=(A1c*1.84)-4.76

As you say, we're all different, have different glycation rates which vary with age (increase with age), have different peaks between readings etc; so I tried all three on my model, using regression, and none fitted! What consistently fitted was using the Nathan formula on my TRUE average derived from my spreadsheet using mean values under BG curves after meals and overnight. It gave me a predicted HbA1c within 0.1% on two separate calculation periods against lab HbA1c results.
So that's what I use and hence my comments about finding proper averages.

Have I saved myself now? :lol:
 
Grazer: that formula isn't really an ADA one.The ADA, EASD and IDF sponsored a study (2006-2008) which was designed "to better define the mathematical relationship between HbA1c and average glucose."
The UK decided in 2008 not to report estimated glucose levels
http://www.acb.org.uk/docs/eAGmeetingreportfinal.pdf

Interestingly enough,I used to use SI diary to log my results. This calculates an estimated HbA1c. (no idea how) It was always spot on; that's on six basic tests: waking, 2 hours after breakfast, before and after lunch, before dinner and bedtime plus some tests at 2-3am and tests during exercise and tests when hypo.
 
Phoenix, the formula did come from that study which included the ADA though didn't it? And which formula is the one you recently suggested was more accurate and what was that, and where from?
Fact is, trying to convert BG into HbA1c is like converting US dollars to a bucket of sand- they are different things and it doesn't work. Seems to me that the best you can do is either use them for the different things they DO represent, or find your own formula which works for you.
 
Back
Top