Bg machine testing accuracy - how accurate

[Applenerd81]

A recent trip to my diabetes clinic for my annual checkup (hba1c 46 /6.3) has led me question the accuracy of my current meter so I thought I'd ask to find out the consensus on how accurate they should be. The following photos show 3 glucose meters and their readings with blood taken at the same time. The variance appears to be near enough 1.5 mmol?! I feel my current meter (Cellnovo) may be under reading.. This tied in with a test I conducted using the hospitals bg machine also.

 

[tim2000s]

This way madness lies.

I'd stop worrying about it. Very rarely do two meters provide the same reading from different drops of blood. Diabetes management is an exercise in dealing with averages and inaccuracy. It doesn't matter what the level is too much. If you are at 5 or 6 then you are in a good range.

It's really not worth spending your time worrying over.

 

[Squire Fulwood]

Meters are inaccurate mostly. The specification says something like they can be out by +/- 15% or worse. I came across one particular make that sacrificed the -15% and added it to the over reading since it always over read. Find one you like, learn its idiosyncrasies and stay with it.

Comparing meters and making allowances for them will keep you occupied for a long time.

 

[June_C]

I did an experiment just recently with 3 different meters for about a week. I suspected one was reading high and the test proved it. Out of the 2 that were nigh on identical, I now use just one of them and the other two are in the cupboard as spares.

 

[Applenerd81]

This way madness lies.

I'd stop worrying about it. Very rarely do two meters provide the same reading from different drops of blood. Diabetes management is an exercise in dealing with averages and inaccuracy. It doesn't matter what the level is too much. If you are at 5 or 6 then you are in a good range.

It's really not worth spending your time worrying over.

Appreciate what your saying but if the meter reads 3.5 mmol and its actually 4.5/5 mmol thats a big difference and the difference between treating a hypo and not... Tricky when trying to stay low carb.

 

[Applenerd81]

I did an experiment just recently with 3 different meters for about a week. I suspected one was reading high and the test proved it. Out of the 2 that were nigh on identical, I now use just one of them and the other two are in the cupboard as spares.

My only worry is that the duff one in my case is my current pump meter!

 

[tim2000s]

Appreciate what your saying but if the meter reads 3.5 mmol and its actually 4.5/5 mmol thats a big difference and the difference between treating a hypo and not... Tricky when trying to stay low carb.

All meters now have to meet a requirement that they are no more than 15% from the real value, so a value of 3.5 with a real value of 4.5 can't occur.

If you consider the distribution of the errors, it is effectively on a normal bell curve, so there will be very few at 15% (it is outside the three standard deviations and is therefore 0.1% of readings), while 68% of readings will be within one standard deviation, which is likely to be somewhere around the 7% level.

 

[Applenerd81]

All meters now have to meet a requirement that they are no more than 15% from the real value, so a value of 3.5 with a real value of 4.5 can't occur.

If you consider the distribution of the errors, it is effectively on a normal bell curve, so there will be very few at 15% (it is outside the three standard deviations and is therefore 0.1% of readings), while 68% of readings will be within one standard deviation, which is likely to be somewhere around the 7% level.

Thanks Tim, I think it would be worth a control solution check at least, especially for lower end.

The readings shown where taken within 5 minutes. Assuming the real reading is 6, my current handset has read between 5.0-6.1mmol - so this is roughly within 2-16%. I'm happy with 2% but 16% is quiet a lot. I guess I should probably do a few more tests if planning on bolusing/treating a hypo..

 

[tim2000s]

Assuming the real reading is 6, my current handset has read between 5.0-6.1mmol - so this is roughly within 2-16%.

That's the issue. You can't assume that the real reading is 6. It could be 5.5 and all the meters would still be correct. What you can at least say from all of those readings is that your blood glucose levels were rising...

 

[Applenerd81]

Think this is a 1-0 to Diabetes!

 

[LucySW]

This way madness lies.

I'd stop worrying about it. Very rarely do two meters provide the same reading from different drops of blood. Diabetes management is an exercise in dealing with averages and inaccuracy. It doesn't matter what the level is too much. If you are at 5 or 6 then you are in a good range.

It's really not worth spending your time worrying over.

Totally, totally agree.

I faffed around with thus, then stopped and got on w analysing my trend data.

 

[Lee Riley]

A recent trip to my diabetes clinic for my annual checkup (hba1c 46 /6.3) has led me question the accuracy of my current meter so I thought I'd ask to find out the consensus on how accurate they should be. The following photos show 3 glucose meters and their readings with blood taken at the same time. The variance appears to be near enough 1.5 mmol?! I feel my current meter (Cellnovo) may be under reading.. This tied in with a test I conducted using the hospitals bg machine also.
View attachment 16204View attachment 16205
View attachment 16207

You need to do more than one test per machine to judge accuracy otherwise it is an inaccurate result in itself.

Note that all meters must conform to ISO standards so will be pretty **** accurate unless one of your meters in particular is very old.

Obviously for test comparisons to be fair, you'd need to change the lancet for every test. Wash/wipe your fingers and dry them per test and ideally do 5 or so tests per machine. Oh and all machines would need fresh batteries. 1.5 is a huge difference but if your true result lies in the middle, then the scale is only within 0.7 of the range and is thus perfectly acceptable.


The leaflet that comes with the meters will tell you about their accuracy.

Any meter that lies smack bang in the middle when compared to others is probably the best bet when weighing up their comparisons and the information provided in each meter manual/leaflet.

 

[Lee Riley]

Read the ISO guidelines for glucose meters. For those of you get mad, don't, unless you wish to make your own meter.

 

[dano17439]

Im close to chucking my machine in the bin. Tested yesterday just before dinner....reading was 33.2!. Pricked another finger straight after and it read 18.1, so done a third test straight after and it was 5.9?

Is it faulty???

 

[dannyw]

Im close to chucking my machine in the bin. Tested yesterday just before dinner....reading was 33.2!. Pricked another finger straight after and it read 18.1, so done a third test straight after and it was 5.9?

Is it faulty???

BG meters rarely play up but it could be faulty. A more probable explanation would be that your finger was contaminated. Always wash your hand prior to testing ( if poss ), but certainly after a reading which you may think is odd.

 

[Robbity]

@Dano 17439: Did you have clean hands?
I just wash or clean my hands and test again with same finger if I get weird readings. Look for possible causes e.g something on your skin - mine are often caused by a bit of residual hand cream - I have one (aloe vera baby cream) that actually tastes sweet. Other things too such the food you've eaten or the state of your health can also affect levels.

But you'll almost always get some (not necessarily huge through) differences if you sequentially test using different fingers: your meter is not going to be 100% accurate (see previous posts) , and your blood's circulating all the time and your glucose levels will be fluctuating with it, so don't expect to get identical readings. Just don't necessarily blame your meter for every unexpected result!

Robbity