Reliability of mySugr hba1c estimation

[everydayupsanddowns]

If you have Libre data, why care much about HbA1c? It's only purpose is to give a rough estimate of a summary of some of the kind of data you get from a Libre.

Yes HbA1c is a flawed proxy measure really. And I find Time In Range significantly more helpful and informative. They can already see from studies that two people with the same HbA1c, but very different glucose profiles ( eg one person gently pootling along and the other veering erratically from high to low and back again), can have very different levels of risk of developing complications from their diabetes.

But at my vintage of diabetes-years, there’s something about the A1c that feels like a grade from an ‘external examiner’, while sensor data has the feel of a ‘teacher assessment’. Ridiculous, I know. But I can’t seem to shake it.

 

[everydayupsanddowns]

As I understand it, HbA1c is a function of how long the patient exists in a state where his/her blood glucose (as measured in a lab environment) is elevated above a defined value (from memory, I think the trigger value is 8 mmol/L, but I could be wrong).

I’m not sure I’m aware of any cut-off value. My understanding is more that it’s just exposure to glucose in the blood. If there’s more glucose in the blood, there’s more chance of a red blood cell getting it ‘stuck to it’ (glycosylated). At lower BG levels there’s must not so much sloshing about for the red blood cells to get affected by. Higher BG levels create an environment where there’s more opportunity for red blood cells to get changed by the glucose they are bumping into. But I thought it was more of a continuum, rather than only happening above a certain point.

 

[littlevoice359]

Now I think of it, I misspoke in terms of a cutoff value for HbA1c - apologies! I was probably thinking of something the tech told me the last time I had my feet checked. If memory serves, he said something to the effect that consistent BG values over 8mmol/L were related to onset of neuropathy. Does that make sense?
Honestly, I'm just figuring this stuff out as I go along, so it's best to take anything I say with a large pinch of salt.
Personally, I start to feel a little off when my BG goes over 8 or under 5, so I try to keep within those two values as much as possible.

 

[Leadinglights]

Now I think of it, I misspoke in terms of a cutoff value for HbA1c - apologies! I was probably thinking of something the tech told me the last time I had my feet checked. If memory serves, he said something to the effect that consistent BG values over 8mmol/L were related to onset of neuropathy. Does that make sense?
Honestly, I'm just figuring this stuff out as I go along, so it's best to take anything I say with a large pinch of salt.
Personally, I start to feel a little off when my BG goes over 8 or under 5, so I try to keep within those two values as much as possible.

I think you are getting confused between the HbA1C test which gives the average blood glucose over the previous three months and is given in mmol/mol. Anything over 47mmol/mol being diabetic.
The numbers of between 5 and 8 mmol/l are results you get from a finger prick and represent the glucose level in the blood sample at that moment in time. People with Type 2 would be aiming at 4-7 mmol/l fasting and before meals and no more than 8-8.5mmol/l 2 hours post meal.

 

[Docb]

But at my vintage of diabetes-years, there’s something about the A1c that feels like a grade from an ‘external examiner’, while sensor data has the feel of a ‘teacher assessment’. Ridiculous, I know. But I can’t seem to shake it.

I rather like that @everydayupsanddowns.

The important point is that the chances of you getting an A in an exam ain't good if your teacher never gives you more than 5/10 for anything you do. In the same way (but upside down) chances are you won't get an HbA1c under 48mmol/mol if all your finger bodges are over 10mmol/l.

 

[Eddy Edson]

But at my vintage of diabetes-years, there’s something about the A1c that feels like a grade from an ‘external examiner’, while sensor data has the feel of a ‘teacher assessment’. Ridiculous, I know. But I can’t seem to shake it.

I guess I'd probably feel the same way if I was a user of CGM of some variety.

Also, I guess I'd want to have a timeline of comparable data starting from before I started using CGM.

But I think there's areas where the difference becomes more significant. I see plenty of research studies where HbA1c is treated as an actual measure of BG. For example, studies pupporting to show elevated complication risk at "pre diabetic" or even high "normal" levels, but which use HbA1c as the measure, without any apparent consideration of the variability. This is face-value flawed, just because it is fairly common for somebody to have non-diabetic HbA1c but average BG above 7.8 mmol/l, ie at diabetic levels.

And once you get past the initial foundational studies establishing the empirical, conventional BG / HbA1c association, the correlation in more recent studies appears to be even lower. This might be an outlier:


buried in the supplementary data of a recent study of "normal" subjects: https://forum.diabetes.org.uk/boards/threads/bg-profiles-in-healthy-non-diabetics.99521/

But it's a respected, highly-cited study, and as it says:

Overall, the correlation was weak (r = 0.27; 95% confidence interval, 0.12 to 0.41) between mean glucose concentration and HbA1c levels, and there was a wide spread of HbA1c levels for a given mean glucose concentration.

How much do you want to hang on a proxy measure with a correlation of just 0.27 with the thing you want to estimate? Bring on CGM for eveybody who needs to measure BG at a good level of accuracy: ie T1's, insulin using T2's, T2's while they are uncontrolled (not so much "normal", "pre diabetic" or reasonably well controlled non-insulin using T2 people).

 

[Docb]

As I have posted before, the graph above is what you get from one individual who has done a lot of testing where there is very good correlation! The paper you have cited @Eddy Edson shows that it is not wise to take one person's data and assume it has universal application.

 

[everydayupsanddowns]

The paper you have cited @Eddy Edson shows that it is not wise to take one person's data and assume it has universal application.

Inevitable I guess there’s gonna be a lot of variation depending on the nature of the completeness of the BG checking routine between individuals, not tomention the nature of what happens to BG ‘between the dots’.

 

[Eddy Edson]

View attachment 22418

As I have posted before, the graph above is what you get from one individual who has done a lot of testing where there is very good correlation! The paper you have cited @Eddy Edson shows that it is not wise to take one person's data and assume it has universal application.

On the other hand, the study in the paper was done with "early generation" Dexcom CGM devices, so maybe there's a lot of individual sensor variances screwing up the correlation.

But on the other other hand, the foundational studies were also done with early generation devices ...

So maybe it's all too hard and the best thing is to say sod it & go for a walk in the park & play with the cockies 🙂