One of my meters is lying

[Oldvatr]

Found an interesting trial report on glucometer accuracy. 4 standard off-the-shelf meters were tested against a lab analyzer and against ISO 2013 levels. Their findings are interesting to say the least. One of the trialed meters was the SD Codefree.
https://www.panafrican-med-journal.com/content/article/32/118/full/

Note this was a trial that was not looking at the possible effects I am exploring in this thread but is a starting point toward a baseline. They found all the trial meters read higher than the true value, and the SD was particularly poor at the hypo detection level it seems.

 

[Oldvatr]

The technology used by SD is Glucose Oxidase (300 units) with potassium ferricyanide reagent.

The meter is apparently sensitive to the following (not a complete list)
Heparin over 3,000U/l
Ibuprofen 50mg/dL
Levodopa 4 mg/dL
Maltose 60mg/dL
Trigs 1.06 mg/dL
Uric Acid 9 mg/dL
Ascorbic acid 4 mg/dL
TC 506 mg/dL
Galactose 60 mg/dL
Hemoglobin 200mg/dL
Bilrubin 40mg/dL

I cannot find any mention of hematocrit sensitivity, but there is a mention of some unnamed parameter to be in the range 20-60% which I think might be the one.

So it looks like the meter may not be suitable to anyone suffering from Parkinson's or kidney failure or laking large quantities of either pain relief NSAIDs or vitamin C. Sodium salicylate is also a noted interferant but not sure if this is Aspirin.

 

[LittleGreyCat]

@Oldvatr Thanks for the thread.
I am also chasing anaemia and causes/consequences.

I don't have the spare mental bandwidth to work through all the information at the moment, but I am following the thread.

I am chasing differences between HbA1c and long term interstitial fluid readings, so I am now wondering if the effects that you are observing and researching could also apply to Freestyle Libre.

 

[Oldvatr]

The technology used in the Caresense Dual is FAD Glucose Dehydrogenase (3.4 units) with an unspecified moderator.

The meter is apparently sensitive to the following (not a complete list)
Paracetamol - Yes but amount not specified.
Uric Acid Yes but amount not specified
Ascorbic acid Yes but amount not specified

Hematocrit sensitivity 15 - 65%

Declaration made that it meets the requirements of ISO 15197-2013 tests for interference and hematocrit.
These are the requirements for ISO 15197-2013 for interference testing to the ISO apparently
https://journals.sagepub.com/doi/pdf/10.1177/1932296818821389

But Caresens do not provide any values in their literature so must assume that their declaration of meeting the ISO is sufficient proof that they have tested these contaminant limits.

Heparin - not mentioned
Ibuprofen - not mentioned
Levodopa - not mentioned
Maltose - not mentioned
Trigs - not mentioned
TC - not mentioned
Galactose - not mentioned
Hemoglobin - not mentioned
Bilirubin - not mentioned

So it looks like the meter may not be suitable for anyone laking large quantities of either Paracetamol for pain relief or vitamin C. Also poor kidney function?

Test evaluation report follows. I think the Caresense Dual is Meter C in this study, but it is not clear.
https://journals.sagepub.com/doi/pdf/10.1177/1932296818821389

Only one batch of test strips was used in the trial

 

[Oldvatr]

I have managed to extract a graph from Excel for the 2hrPP against the Pre readings. i.e. a delta due to eating a meal

The readings do show occasional drops and this seems to be where I had a high PRE due to a snack in the day, and shows my Stage 1 Insulin response at work. Most of the time, the SD reads higher.

 

[Ronancastled]

If they just made a test solution that was say 5.5 mmol/L +/-0.1 then we could check ourselves.

 

[Oldvatr]

If they just made a test solution that was say 5.5 mmol/L +/-0.1 then we could check ourselves.

I totally agree. Or print an accurate assayed value on the vial. But I suppose over time it might become contaminated, or dehydrated thus affecting that accuracy uncontrollably. However, that would not be representative of blood values and blood contains lots of funny compounds that the test solution does not include and which affect readings. So hematocrit changes daily and the enzyme is affected by glucose in the RBC too as well as PO2 oxygen levels and lots of crud like bilirubin. Proper witches brew that a simple sugar solution cannot emulate. That is why the current test solution is a simple working/ not working test, not a calibration.

 

[Oldvatr]

Came across a paper which describes the technology used in electrochemical biosensors.

It describes the technologies used in the two meters we are discussing here and was published in 2020. It also covers recent advances in non-invasive and CGM sensor technologies.

Definitely for the Nerds only, but informative. The pdf file is too large to upload to a post but the abstract is available here
https://pubmed.ncbi.nlm.nih.gov/33113948

 

[Bubbleblower]

OK done a bit more research, and found the following reports which are of interest
https://care.diabetesjournals.org/content/26/12/3215
and
https://diabetesed.net/how-accurate-are-glucose-meters/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2769960/
https://agamatrix.com/blog/blood-glucose-meter-accuracy/

Unfortunately IRL blood sugar meters aren't that accurate and it changes every batch of teststrips. One of my meters is the same brand you have and it deviates upto +50% compared to lab tests after only a small sample. My hematocrit levels are very low.

One of the studies you qouted found errors up to 43% and the study "Blood glucose meters Performance of devices on the Dutch market" found 44% and 3 out of 4 meters scored insufficient on analytical performance, mainly due to non-compliance to requirements with regard to interfering substances.

 

[Oldvatr]

The effect I am researching is a dynamic effect that is short-lived and seems unconnected to the meter's accuracy or gain factors. It is directly triggered by eating a meal and is therefore probably related to an interference issue that one of the two meters is more sensitive to than the other. It must be bloodborne, so something in my blood changes following the meal, and drops back to baseline a few hours after. Whatever the substance is, the effect seems to be changing over a long timescale, so is probably not related to meal content. I strongly suspect that it is something going on in my body that is changing over time. If I can isolate the substance that the meter is reacting to, then I will have a better idea of what is changing in my body, and whether it is something of concern in the future.

For instance, a change in hematocrit could indicate anemia worsening, or an infection, or cancer, say.
A sensitivity to urea levels could indicate a kidney problem. A change to Bilirubin would indicate a gall bladder or bile duct issue.

A change in ferritin could be a diet deficiency or anemia ( but this is not mealtime related, so ignore).

There are many other things I could list, but those are the ones that spring to mind as I write. It is probably not Maltose or Galactose since that would be meal content dependant. It could be lipids or Trigs increasing which would tend to point to the gut microbiome, enzymes, or stomach acid changes.

I note that the list of interferents does not include sodium or potassium.

I am negotiating a blood test from my GP to see how it has changed since 2019 but still waiting for a response.

 

[Bubbleblower]

If I can isolate the substance that the meter is reacting to, then I will have a better idea of what is changing in my body, and whether it is something of concern in the future.

That would be quite an achievement if you manage to determine what causes this.

The problem I have with my BGM seems unrelated; the results from my Caresens are almost always higher and the effect is also preprandial. The readings change rapidly, for example 2.5 points within a minute. My nurse uses the same meter and according to that my FBS was 13.5 shortly after diagnosis. A week later my own BGM arrived and my FBS was supposively 20.6.

My other meters that don't have this problem use the same GDH-FAD enzyme.
Fortunately they keep each other honest and improve accuracy, especially if they are not so accurate.

 

[Oldvatr]

That would be quite an achievement if you manage to determine what causes this.

The problem I have with my BGM seems unrelated; the results from my Caresens are almost always higher and the effect is also preprandial. The readings change rapidly, for example 2.5 points within a minute. My nurse uses the same meter and according to that my FBS was 13.5 shortly after diagnosis. A week later my own BGM arrived and my FBS was supposively 20.6.

My other meters that don't have this problem use the same GDH-FAD enzyme.
Fortunately they keep each other honest and improve accuracy, especially if they are not so accurate.

It is clear from my experience that you must discard the first drop, then squeeze the next one gently. The first drop is stale interstitial fluid, and not indicative of the current blood reading. Squeezing too vigorously forces more interstitial fluid from surrounding tissues, thus affecting the capillary blood which is what we want to measure. My other experience is that it is important to wash or clean the test area and that using some soaps, detergents, or spirit-based wipes can also induce errors. I make sure I use the same soap and use lots of warm water before using a clean cloth to dry to try to make all my readings repeatable. Mind you, the meters themselves have uncertainty that will always introduce some jitter or noise naturally.

If I average the differences between all my pre-readings, then they seem to cancel out between meters,so that is how I know that whatever offset value I get, it seems to be constant and not affected.by gain or reading value.

 

[Antje77]

It is clear from my experience that you must discard the first drop, then squeeze the next one gently. The first drop is stale interstitial fluid, and not indicative of the current blood reading.

I usually get the same result with first or second drop. I've tested this because I was always using the second drop and read somewhere it didn't matter.
I wanted to see for myself and now I use the first drop, allowing for a lower setting of the pricker.

 

[Oldvatr]

I have had a chance to analyse the data with respect to the post meal spike, and I have to admit that I cannot see any correlation between an error increase (delta) from the pre-meal reading of more than 1.0 mmol/l, which is roughly what the ISO standard says can be the reading error allowance allowed. Although there were meals when this error was excessive, there were also days where the error was within tolerance for the same meal contents. For example, a salmon steak meal will be 2.2 on one day but only 0.3 on a subsequent day. Sometimes, my 2hr PP would drop to being lower than my pre-meal.

So I have to declare it as having no apparent correlation to what was eaten in the meals.

So I have to declare my SD Code-free is lying and seems to be giving errors outside the scope of the ISO requirement. It seems to be a random error but may be a test method problem i.e. butterfingers. I do sometimes do a retest, and my Dual seems fairly consistent, but the Codefree still swings around like a windsock in a gale. My weighing scales do the same thing. First reading = high, Step off/on again and the repeat is lower, but will stick with the new value no matter how much I abuse the scales after.

 

[LittleGreyCat]

First reading = high, Step off/on again and the repeat is lower, but will stick with the new value no matter how much I abuse the scales after.

That triggers a memory of sorts.

I think that when using laboratory scales (Oertling balance, which always reminded me of a hurtling balance) you were supposed to make the balance move to remove any stickiness from being at rest before doing the real weighing.

It suggests that your scales might have a tight spot which needs easing before an accurate reading.

Then again, a long time since I did any real science.

 

[Oldvatr]

OK. I have only glanced at this paper, but it looks promising.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2851212/

I am looking for something that changes after a meal but drops back after about 6 hours. Now two spring immediately to mind. Insulin and C-peptide. I doubt my SD is sensitive to insulin concentrations, but recent research into c-peptide has established that it is an active participant and is not inert. It used to be considered that it just joined the A and B molecules of insulin when released into the bloodstream so that insulin becomes active by both molecule chains joining together instead. And that hence the c-peptide bridge was just a discard. But it seems it does play further roles in metabolism in its own right.
https://diabetes.diabetesjournals.org/content/61/4/761

One aspect I am sensing as a possible interferent is the reported interactions at the cell level. I mean, what is an enzyme when it's at home? It's cells that react biologically to certain biological reactions and create other chemical markers such as hydrogen peroxide that can be detected. Maybe the specific enzyme used in the SD is being corrupted by c-peptide in some way? I am now out of my comfort zone at understanding the chemistry since I only did A level. So I throw it up in the air in case anyone knows a better-qualified personage than moi. But I feel this may be a possibility as the timing is about right.

Parking that for the while, I will look at the other things that vary in the blood after a meal.

 

[Richard'63]

That triggers a memory of sorts.

I think that when using laboratory scales (Oertling balance, which always reminded me of a hurtling balance) you were supposed to make the balance move to remove any stickiness from being at rest before doing the real weighing.

It suggests that your scales might have a tight spot which needs easing before an accurate reading.

Then again, a long time since I did any real science.

We call it stiction

 

[Oldvatr]

I have done some thinking on this c-peptide possibility, and testing my hypothesis may be fairly simple.

Firstly I need to see if the effect is limited to my current meter, or is systemic over more than one meter. So I need to repeat my n=1 experiment with one or more new meters using the strips I have already bought. I will contact Homehealth to see if they can assist me with this.

If it is systemic, then testing for c-peptide requires some volunteers who are insulin dependant and some who are on orals or lifestyle control to do what I am doing - parallel testing. If there is an interaction, then the insulin users should not see much of this effect, but the oral users should. If neither or both cohorts sees the effect then it's not c-peptide after all.

 

[Oldvatr]

Just read the Homehealth sales pitch for the new Navii glucometer
"Upgraded meter from SD Biosensor, the makers of the top-selling SD Codefree Blood Glucose Meter. This new meter has all the benefits of the Codefree but boasts greater accuracy due to a wider haematocrit (HCT) range of 0-70% and GDH-FAD enzyme technology. This means that the results are not impacted by extremes in HCT nor variations in oxygen concentration respectively."

There may be a clue here - maybe do breathing exercises before jabbing yourself and don't hold your breath while squeezing the droplet out of a reluctant finger?

As per the first linked study in my post a couple of days ago, eating a light meal can drop HCT by almost 5%, and my last HCT measured at 38 so could be approaching the nonlinear cutoff region at the lower end. However, looking at the graphs in that study, the HCT value continued dropping after the 2hrPP and the 4 hrPP value was the worst, which is not tying in with the peak error in my readings which was the 2 hr PP mark. So not convinced my observation is HCT related. A high lipid load (HF of LCHF) may skew the result tho.

 

[Antje77]

If it is systemic, then testing for c-peptide requires some volunteers who are insulin dependant and some who are on orals or lifestyle control to do what I am doing - parallel testing.

I'm always in for experiments, but I'm afraid I use different meters than you do: the Caresens N premier and the Contour Next One, so not sure it will be of use to you.
I also may not fit either of your test groups, being insulin dependent but at my last (and only) C-peptide test (2 and a half years ago) I still produced some insulin, which may skew your experiment if I still do, which I have no way of knowing.

Still if you want me to do some testing, just let me know what you'd like me to test