Accuracy of home blood glucose meters

[increasingly cynical]

Hi All,

Does anyone have any information regarding the accuracy of home glucose monitors (I use acu-check aviva) because:

The details on the packets of control solutions you use to calibrate the meters every now and again give a tolerance range for level 1 solutions of 1.4-3.1 and for level 2 solutions 14.2 -19.1.

The difference in the tolerance ranges for the two control solutions suggests that the meter is expected to be more accurate at lower readings and less accurate at higher readings (which might be expected as very low levels are more immediately life threatening than very high levels).

I called the acu-check helpline to check I had understood this correctly. They weren’t able to answer the query about meter accuracy directly, but they did say that for lower readings they are happy to accept an error margin of 15% and at higher blood glucose readings they are happy to accept a margin of error of 30% (!!). They say this level of accuracy meets ‘MHRA’ standards (whatever these are).

I find it quite disturbing that this might be the range of accuracy of your average meter. For example, at higher readings, it could be the difference between you assuming that you have a BG of 14 (or of 26 for that matter) when actually, your ‘real’ reading is 20. This could be the difference between someone deciding to take themselves off to hospital and deciding they were high but OK!

I forgot to ask them if the error margins are expected to be fixed or random. If an individual meter always makes a mistake in the same direction, then you can control for that, but if it makes errors in both directions then I would have thought that using a meter to judge what food you should be eating becomes something of a gamble (for example, your ‘real’ reading is 14, but the machine told you it was 9.8..you adjust your food accordingly… your next reading is really 14 again (let’s say) but this time the meter tells you it is 18.2 and you begin to panic…

Anyone have any thoughts or experience of the above? Was the helpline guy right… was he wrong… is there something I haven’t understood about meters and how one uses them in controlling diabetes?

Thanks!


:shock:

 

[Dennis]

Hi IC,

I've tried meters that need regular calibrating with a control solution and have found the ones I used to be nowhere near as accurate or consistent as the one that I now have that never needs calibrating. MHRA is the Medicines and Healthcare Regulatory Agency - the body that lays out the standards that health product manufacturers are supposed to follow.

All meters are supposed to operate within a tolerance of +/- 15% and you are right that this gives a margin for error. However, a meter is supposed to be used as a guide, not give you results that are guaranteed 100% accurate. When my meter says 6.0mmol, I accept that this might actually be anything between 5.1 or 6.9, and with either of those I am happy.

 

[increasingly cynical]

Hi Dennis,

Thanks very much for your reply and for info on what the 'MHRA' is. In terms of 'calibrating' i didn't mean that my meter needs to be constantly re-calibrated to work, just that you have to check against control solutions every now and again. I don't know ho wthe 'acu-check aviva compares with other meters in terms of accuracy.

I still find a 15% error margin unnaceptable, but the 30% at the high end worries me more.. 30% of a high number is a lot ... a difference of 6 in BG readings equates to an awful lot of food or exercise and to an insulin dose that could give you trouble if the meter was wrong!

I wonder if there is a way of checking the margin by which any individual meter is out?

 

[Dennis]

In terms of 'calibrating' i didn't mean that my meter needs to be constantly re-calibrated to work, just that you have to check against control solutions every now and again.

Hi IC,
It's the control solution check that I meant by recalibrating. My present machine never needs this.

I wonder if there is a way of checking the margin by which any individual meter is out?

I believe that some manufacturers will send you a bottle of test solution that you can use to check the meter's accuracy. This has a known sugar content so should always give a consistent answer when tested. I don't know if all the manufacturers offer this.

 

[increasingly cynical]

Hi Dennis,

Thanks again for your replies. That was very helpful. Thanks also fro mentioning the freebie control solutions, accu-check are now sending me some of these. Funnily enough, I had a call back from the accu-check careline tonight, the person I spoke to earlier in the day was a trainee apparently and hadn't got the information exactly right. The person who called me back gave me a lot more detail and some of it is quite useful info, so I thought I would copy what he said here in case anyone else is interested:

Accu-Check Aviva Meter Careline made the following statements re meter accuracy (3rd Feb ’09)

The MHRA allow a margin of error of 15% across the standard scale from 2.8-33.6

At the lower end of the scale, which runs from 0.6-2.7 the allowable margin is an absolute value not a percentage. The figure is 0.8 mmol/l (a margin of error of more than 100% on 0.6 and of 28% at 2.7).

The nature of variations in an individual meter is difficult to predict, there is some expectation that they would be more likely to be in one direction only, but they could also vary in both directions.

Whilst the control solutions give a range for both Level 1 (low BG) and Level 2 (high BG) controls, this is to take into account deterioration of both test strips and control solutions after their respective containers have been opened.

If both test strips and control solutions come from newly opened bottles (opened immediately before you calibrate your meter) then, the value of each control solution in mmol/l should be exactly in the middle of the relevant range.

The above point means that, by using newly opened test strips and control solutions, you can check the specific ‘margin of error’ for your own particular meter. Testing it several times over a fairly short period (so that the strips and controls are still fresh) will give an idea of whether there is any ‘random’ variation (if there is a lot of random variation its probably best just to get rid of the meter).

8)