My Meter Readings

Which brings me back to the question of why your machine is expected to read control solution as if it were blood glucose, showing the actual concentration, while mine sets a completely different range of tolerances for the control (?)

Ah, sorry, you calculated those tolerances yourself.

This is a bit confusing. If you calculated them yourself, then it could be that the control solution is expected to read high, like mine, so you might not be able to infer from your results that the meter reads high for blood glucose.

Or did you calculate your own control tolerances?? I've lost the plot now! Please summarise!

Yes, so my comment stands. If, as with my machine, control readings are expected to be systematically high (because it is not real blood), the same expectation should hold for your machine too. So for a concentration of 2.8 it might be correct for you to get 3.5 on the control test. The real blood readings will then be lower, as are mine.

So this is really interesting.

It seems from all of your available data that the SD machine reads high for blood by about 1.0, but also that it reads high by the same margin for the control solution!

Implication being that meters should be expected to read control solutions more or less as if they were real blood - rather than with a systematic bias because it is not real blood.

But mine appears to expect a systematic bias. (??)

Haven't heard back from Roche yet.

Yes - so if all of that is correct I am left with a mystery:

Why do the manufacturers specify that my control solution should read about 8.0 (and indeed it does) if it is actually 6.7?

Going back to the hospital results, they showed 5.3 against my own meter's 6.3 - 6.7. That would tie in with an error of 8.1 rather than 6.7.

If everything you have said is correct, there seems to be something wrong here.

The manufacturer specifies a range between 6.7 and 9, from which I infer that the correct reading for this machine and solution ought to be about 8.0, thus allowing an equal margin in both directions.

If I am correct about this, the mystery is why they expect such a high reading, when all of your evidence suggests that control solution ought to come out reading as the real concentration. It would surely be because the machine is set to read everything high.

If I am wrong about this, then your speculation about setting the control solution error margin artificially high looks right. But this seems unlikely; If I were to get a reading of 9.0, that would mean the machine is out by about 34%, which is not permissible. In addition, it really does read as 8.1, so even then the machine is reading high.

It looks as if it is reading high whatever assumptions we make! Simply because all of your evidence leads to the conclusion that control solution would normally read accurately as real blood.

Here's their response - just in. I can't see anything about a systematic error expected in control solution (?)

Thank you for contacting Accu-Chek.

Please be aware that performing a control test in your home environment cannot by compared to performing a control test in a laboratory, as the strict laboratory testing conditions are not met in the home environment. This means that achieving the target reading of 6.67 mmol/l is not possible. Due to these factors a target range has instead been calculated. This range takes into consideration the effects of temperature deviations and possible sources of contamination. The specific target range for the Accu-Chek Compact Plus control solution is printed on the peel-off label attached to the drum container.

Please see below important information about performing the control solution test:

Treating the control solution properly:
The solution is filled into the bottle under microbiologically controlled (low germ) conditions. To avoid contaminating the solution, do not touch the tip of the bottle with your fingers, the test strips or the surface onto which you apply the control solution. Contaminated control solution can produce incorrect results.
Use only control solution that is within its expiry date. Expired control solution can produce incorrect results. The expiry date is printed on the bottle.
Store the control solution between +2 and +30°C. Control solution that is stored outside this range may become unusable before it reaches its expiry date.
If you store and handle the control solution properly, it will remain stable for 2 months after opening. However, do not use control solution that is past its expiry date, which is printed on the bottle.
The control solution must be at the ambient temperature when you perform a control test. If you keep the control solution in a refrigerator, leave it to stand at ambient temperature before you perform a control test. Cold control solution can produce incorrect results.

Please also note:
Reproducibility (day-to-day imprecision)
The mean imprecision is < 2%. In a typical series of tests a standard deviation of 0.15 mmol/l was obtained with a glucose concentration of 8.6 mmol/l.
Calibration
The system (meter and test strips) is calibrated with venous blood containing various glucose concentrations. The reference values are obtained using the hexokinase method. This reference method is traceable to a NIST standard by means of the ID-GCMS method, the method with the highest metrological quality.


If you have any further questions please do not hesitate to contact us.

Kind regards,

Accu-Chek Customer Service Team

My thoughts precisely.

If I understand them correctly, they say that because of temperature variations and possible impurities it is not possible to obtain the perfect reading (which they imply would be 6.7) for a 6.7 solution. So, as you speculated, they have artificially raised the target range to be on the safe side. My 8.1 reading merely serves to demonstrate that these distortions occur.

Yes?

So then why would these same temperature variations and possible impurities not affect real blood glucose readings in the same way?

So whatever the explanation, in the absence of an exaggerating effect in control solution but not blood, the meter does read high by about 1.3.

My reply to them:

"Thank you very much for the detailed reply.

I do want to assure you that I do not intend to extend this exchange beyond what is necessary to clarify the relevant point for me, but I am unfortunately still unclear. My question now is:

You seem to be saying that because of home testing uncertainties, the control solution is likely to read high. Indeed it does, on my machine, at 8.1 instead of 6.7 mmol/l.

If this is correct, then is the machine equally likely to read high with actual blood serum? If my machine reads my blood as 8.1 mmol/l, does that suggest that it is in fact around 6.7 mmol/l?

I need to know this so that I can interpet my readings correctly.

Kind regards,
etc.."

Update:

Last Friday I received an email from Roche, promising an expert would get back to me on Monday. Still no expert.

I wonder why they are finding this so hard?

Call me Mr Cynical, but the harder they find it to reply coherently, the more tempted I am to infer that the meter does, of practical necessity, read high, and that to state this fact outright is to detract from their image.