Quote:
|
You might want to rethink this one. I think a percentage may be more meaningful such as +/- 20%.
|
it's a good point
Meters are more accurate for the range they are calibrated i.e. below 10 (180). There are however different kinds of accuracy.
1) Accurate to a gold standard. I grabbed my ultrasmart bottle and had a look at the control sample range, which is 5.4 to 7.2mmol/l (98-131) or +/- 0.9 mmol (16.5) or as a %age +/- 16%. This is for a calibrated control sample, made in a lab, and not squeezed out of your finger (with all the sample error problems above). The 20% error number comes from testing meters against the results obtained from venous blood drawn in the clinic. If we assume this is correct then we get +/- 1.3 mmol (23 mg/dl) for normal blood glucose ranges with good sampling technique. Gold standard accuracy is about asking whether my 6.3 is really a 6.3. What we find is that 6.3 means a blood sugar that in reality could be as little as 5.0 to as high as 7.6.
2) Consistency. Are the meter readings consistent? If I measure the sample over and over again do I get the same result? On a calibrated sample meters do very well. Run a calibrated sample over and over again, and your meter will read the same result with a max variation of +/- 0.1 mmol/l (1 mg/dl), which sounds really good. Measure yourself and a big variation suddenly appears sometimes even when you measure the same drop of blood. The poster who had 128 from one finger and 140 from another is looking at a sample variation of about of +/- 10%. Translate this to normal BG's and for your 6.3 you get a range of 5.7 to 6.9, so you can be pretty certain that 8.0 is higher than 6.0, but not so certain 7.0 is higher than 6.0.
I think that's alll as clear as mud
