Posted by on Wednesday 25 April 2012

Meter accuracy - the narrow window

I've been stewing over the question of meter (in)accuracy over the last few weeks after my experiences with the iBGStar. Possibly because throughout much of April my levels have been a bit wobbly which often puts me in a grumpy frame of mind as Stacey at noticed. Not helped by the report today that suggest 80% of the massive NHS diabetes budget goes on treating (largely avoidable) complications.

In our fight to avoid complications our 'window on the world' of how our choices are affecting our bodies are blood glucose (BG) meters. BG meters currently have to meet accuracy criteria set by the International Standards Organisation in order to pass muster. This is a good thing.


According to ISO15197 the required level of accuracy for BG meters is +/-20% against lab results 95% of the time. So the very best they need to aim for is that a lab reading of 7.0mmol/L (126) would show as somewhere between 5.6 (100) and 8.4 (151). And 5 times out of 100 it could be much worse. In other words testing at 5.6 and then 8.4 just 5 minutes apart using the same meter with the same pot of strips (or even seconds apart from the same drop of blood!) would pass the International Standards Organisation criteria for accuracy.

Similarly under the requirements 3.2 (58) and 4.8 (86) could be identical readings from a lab test of 4.0 (72). But 3.2 and 4.8 are not the same as far as anyone trying to manage their diabetes is concerned. One is 'reach for the jelly babies' time, the other is 'nothing to see here', while the actual result was 'better keep an eye on that'.

With modern diabetes management techniques we have the potential to manage our condition with a level of accuracy which should guard against complications. But the snazziest insulin pump and whooshiest rapid analogue insulins are being let down by a sloppy data feed.

To play it safe perhaps I should adjust my pre-meal targets to take account of potentially errant readings. So lets see... with a 20% buffer at either end that gives me a target 4.8 - 5.6 (86-100). Aim for that narrow a window? Hmmmm, perhaps not. Unless I want to drive myself mad.

Fortunately most meters far exceed the required level of accuracy most of the time. But as for the 'official' requirements... Good enough? No I don’t think so really.


  1. I've had that problem for a while, and it was especially noticeable when I was on the Continuous Glucose Monitoring last month - the readings were so different. With my BG generally though, I tend to take into account 'feeling' when I see a reading - if for example it says I'm high, but I feel fine, I'll always test again after really washing my hands thoroughly (although I do that anyway) and will never do too big a correction bolus. It is frustrating but hey - that pretty much is life with Diabetes! Glad I read this :)

  2. Well said.

    I think we need two simple things:

    1) BG systems need to meet existing standards in the real world. (aka post market) They don't. What good are tighter standards if the existing ones are not met?

    2) Tighter standards and the process is inspected so it delivers what we expect, Trust but verify.