phoenix
Expert
- Messages
- 5,671
- Type of diabetes
- Type 1
- Treatment type
- Pump
Meter accuracy often comes up. Unfortunately the expected standards of accuracy aren't very high
Monitors with a CE mark should fulfil the following standard:
95% of results should be:
within + or - 15mg/dl at glucose levels less than 75mg/dL
within 20% at BG concentration equal or greater than 75mg/dL.
(15mg/dl = 0.83mmol/l, 75mg/dl= 4.16)
So for 'hypo' levels they should be within 0.8mmo/l and above that should be within 20%
A study earlier in the year found that only 16 out of 27 monitors actually fulfilled the standard .. Looking at the results I don't recognise the names of most of the meters that failed, whereas I do recognise many of those that passed . However they don't surpass the standard by much. The American FDA is considering raising the standards. and If it were raised to 95% of tests within +/- 15% then none of the meters would pass.
The paper also analysed all the errors using a Clarke Error Grid.
http://en.wikipedia.org/wiki/Clarke_Error_Grid
This is a way of sorting out how importantt the errors were. There are errors within the 20% tolerance, those that were outside this but wouldn't cause a clinical problem, these are 'acceptable' errors. There are also those that might lead to overtreatment and (in my opinion) very serious errors that would confuse treatment of hypoglycemia for hyperglycemia and vice-versa.
There fortunately weren't that many errors in the last 2 sections and for the most part I haven't heard of the meters which made more than one of these errors. The Avia and the Freestyle lite each had 1 error that might have caused overtreatment . However, these 2 meters were amongst those that passed the present standards (and indeed were amongst the more accurate meters apart from this 1 reading.)
Table 1 shows accuracy figures, as they only seemed to use 1 example of each meter it is not necessarily true of all meters of that make/model. The study was funded by Roche (but other meters do equally well) http://bionimeusa.com/files/journal_pap ... Report.pdf
Monitors with a CE mark should fulfil the following standard:
95% of results should be:
within + or - 15mg/dl at glucose levels less than 75mg/dL
within 20% at BG concentration equal or greater than 75mg/dL.
(15mg/dl = 0.83mmol/l, 75mg/dl= 4.16)
So for 'hypo' levels they should be within 0.8mmo/l and above that should be within 20%
A study earlier in the year found that only 16 out of 27 monitors actually fulfilled the standard .. Looking at the results I don't recognise the names of most of the meters that failed, whereas I do recognise many of those that passed . However they don't surpass the standard by much. The American FDA is considering raising the standards. and If it were raised to 95% of tests within +/- 15% then none of the meters would pass.
The paper also analysed all the errors using a Clarke Error Grid.
http://en.wikipedia.org/wiki/Clarke_Error_Grid
This is a way of sorting out how importantt the errors were. There are errors within the 20% tolerance, those that were outside this but wouldn't cause a clinical problem, these are 'acceptable' errors. There are also those that might lead to overtreatment and (in my opinion) very serious errors that would confuse treatment of hypoglycemia for hyperglycemia and vice-versa.
There fortunately weren't that many errors in the last 2 sections and for the most part I haven't heard of the meters which made more than one of these errors. The Avia and the Freestyle lite each had 1 error that might have caused overtreatment . However, these 2 meters were amongst those that passed the present standards (and indeed were amongst the more accurate meters apart from this 1 reading.)
Table 1 shows accuracy figures, as they only seemed to use 1 example of each meter it is not necessarily true of all meters of that make/model. The study was funded by Roche (but other meters do equally well) http://bionimeusa.com/files/journal_pap ... Report.pdf
