PDA

View Full Version : A muse on Meters


REDLAN
04-15-2008, 04:13 PM
seems to me that there have been a spate of peeps posting about meter accuracy.

so I thought I would muse on why it is that meters have an accuracy of around +/- 20%.

The first concept to grasp is the idea that there is no such thing as a blood sugar level - rather what we have are blood sugar levels - it will vary depending on which tissue it is measured in. whether we are talking venous, arterial, or capillary blood. level of blood supply, metabolic activity of tissues. In short no 2 fingers even if they are next to each other will have the same BG.

then there are issues with how the blood is collected - they discovered that squeezing causes plasma to appear, lowering the apparent reading. Question how do think depth of penetration of the lancet will change things? How many people use a blunt lancet - tearing of tissues will alter BG reading.

now the meters, and the illusion of accuracy - we get sucked into believing that when our meter says 8.1 mmol or 146 mg/dl that our BG really is this value. Because why would it say 8.1 if it was something else - really we're talking marketing influcences here - what would you prefer a meter that said 8.1 mmol, or one that said 8 mmol?

onto meters...

And what does a meter measure?

blood glucose of course!

actually the one thing that it doesn't measure is blood glucose. What you actually own is a sophisticated voltmeter. On the older style meters (I understand how those one's work, but the principle is the same for the newer meters)

you have a test strip. One surface of it is impregnated with glucose oxidase and a dye The dye for now is colourless. Essentially glucose oxidase breaks down glucose, which then changes the colourless dye to blue.

the strip usually has a very thin gauze strip, which ensures that a set volume of blood is brought into contact with glucose oxidase.

the meter has a little LED which emits light in the red wavelength, and a photoelectric sensor. So...

when you plug the strip in, it knows the strip is there as a voltage will be registered at the photoelectric sensor. When you add blood, the strip will start to turn blue, which will change the voltage registered by the sensor as the amount of light reflected by the strip will change. It knows that you've added blood, because it will flash the led every half a second or so. It then waits a set time, and records the voltage from the sensor (my one did a series of flashes). From this voltage it then applies an algorythm, and then confidently displays....

8.1 mmol on the LCD screen.

so just how many errors could there possibly be? These are ones just off the top of my head...

1) Any variation in blood volume on the test strip will alter the result - increase the blood volume on the strip, and the glucose oxidase has a greater supply of substrate, and will turn more of the dye blue resulting in a higher reading.

2) variation in ambient temperature - for each 10 degree Centrigrade rise in temperature there is a doubling of reaction rate. Test on a hotter day, strip turns bluer and your BG will be apparently higher.

3) timing pulse - remember the pulsing LED... variation in when you put your finger on the test strip, will change the result. give glucose oxidase half a second longer to react it will convert more glucose, and so you will get a higher reading.

4) any variation in battery power will affect the amount of light emitted by the LED, and hence the result. Bit of dirt on the photoelectric cell? same effect.

And why might a high BG be less accurately measured than a low one?

it's all down to glucose concentraion and timing pulses.

the voltage reading is made after a set time. The meter does not wait until all the glucose has been used up, but measures it at a point sometime before.

When you plop the blood on, the reaction rate is relatively high, and then drops as the glucose is used up. For low BG's nearly all of the glucose has been used up by the time the machine makes it's measurement, so any slight differences in timing will make only a very small difference to the reading.

with a high blood sugar, then relatively more glucose will be available, so timing now has a bigger impact as there is more colour change from one second to the next.

PS theses are all my musings - I didn't research them, I just applied a few principles to my understanding of how meters work.

the point is this.

for the amount of money you are prepared to pay for a meter, then it's a miracle quite frankly that they are as accurate as they are!

Even the HBa1c machines give you an A1c which is +/- 0.2 to 0.3.

Scratch
04-15-2008, 04:40 PM
Yeah, people keep seeming to think that the meters should give spot on accurate and precise numbers (**** I wish I could remember better my science classes and the distinction between accuracy and precision) but somewhat indirect methods are being used to get an estimate. Even though the meter spits out a number that our brains like to think of as being a number, like 140 = 140 it's not really that way with the meters. They are bound to be imprecise by the method used along with sampling issues from the individuals acquiring the blood samples.

Quite seriously that people need to be a bit more accepting of a notion of fuzziness.

I also say quite seriously don't give the meter makers and test strip makers any more reason to doing research into more "accurate" meters because they will just charge us more money for that.

The degree of accuracy we have now is sufficient to manage our blood sugars well enough.

In some ways, I think the only real advancement left to be made of importance is a cure. You want something that can really do a nifty good job at measuring blood sugars and administering the correct amount of insulin?

Working beta cells for the type 1s, alleviating insulin resistance for the type 2s.

BlueSky
04-15-2008, 04:41 PM
Thanks for that explanation. There certainly is lots of room for stuff to go wrong. I am not sure how accurate my meter is, and it doesn't really matter. I have learned how to interpret the numbers in terms of what they mean for me. More important than the absolute level is the rate of blood glucose change. So what is most important is consistency between readings. And I am very impressed with my meter (AccuCheck Performa) on that score. Repeat tests are typically within 5% of each other.

JediSkipdogg
04-15-2008, 04:43 PM
Thank you for that post. It should also be noted that for higher accuracy...

A larger sample size is required. Think of it in measuring how many Chinese are in the US. Are you going to measure just the population of San Francisco or the population of the entire US?

With regards to timing, you don't want to rush it but you also don't want to wait until the blood dries up as then it's basically like measuring a rock. However, it generally takes 2 minutes for 1 uL of blood to harden and become unusable. Current meters I believe take one instant reading. Why don't they take multiple readings over 10 seconds and average them out?

The most accurate BG meter on the market today is the Hemocue... HemoCue Worldwide - Products - Glucose - General (http://www.hemocue.com/index.php?page=3005)
So why don't people use it?
-It requires 5 uL of blood. Last I checked people are happy we got away from the 3uL strips and can get samples under 1 uL.
-Results take 40-240 seconds depending on how concentrated the blood sample is with glucose.
-The cuvettes (version of strip) must either be refridgerated or frozen. They only last 3 days at room temperature due to the chemical compound added to obtain the results.
-The meter to start with is $800 and NO insurance covers it.
-The cuvettes are $2.60 each and must be bought in packages of 100 for $260.

The above is what we would pay for precision and accuracy in blood glucose meters.

Now, for a side thought, do we really even care about what the blood glucose level is? CGM devices read interstitual fluid. Interstitual fluid is what is surrounding all the cells of our body. It is how nutrients and waste are transferred between the blood stream and cells. When you start to feel the symptons of a low, it's because your interstitual fluid is low in glucose to give to the cells. That is why one may feel a low before their blood sugar reads it or vice-versa. That is one reason CGM devices have the 10 or so minute delay. Both blood and interstitual fluid have their pros and cons. So which should we be measuring?

notme
04-15-2008, 06:34 PM
Thanks REDLAN, not only was that a GREAT explaination, but I loved the phrase "spate of peeps". ;)

I love this site.

REDLAN
04-16-2008, 05:11 AM
I think the point to take away is that what matters as Bluesky mentioned is that your meter enables you to manage your diabetes effectively.

it is probably not a good idea to use 2 meters - even if they are both the same model, becuse which you have the issue of which one do you trust?

with high blood glucose readings there is a loss of accuracy, and you may wish to consider that "larger sample size" mentioned by JediSkipdogg, and then take the average of your results.

Scrabblechick
04-16-2008, 08:42 AM
That's sort of what I do if I get what I think is an unnaturally high reading. I go and wash my hands and do another test. If it's within 10 points of the other, I resign myself to the fact my Bg is higher than I'd like, but write down the lower number in my log book. Heh.

If I get a dramatically different result, I'll wash again and do another test. Generally, this test will be closer to one extreme or the other. Hopefully, to the lower end. ;)

If I get a result sort of in the middle, I'll average the three and go with the average. Since I'm not on insulin, I can do this.

OH! And thanks for posting your analysis, Redlan! English major that I am, it never would have occurred to me to have looked at my meter in those terms. Your post was informative and very, very helpful!

Razo681
04-16-2008, 11:53 AM
I agree totally on what has been said. Just some comments with my experience so far.

I can get extremely high results if I don't get enough blood on the strip and add more within a 2 seconds (usually high 200s to 300s) when another test will show a normallish range.

What makes me mad is the ratios given to me by my Dr. Granted he isn't a ENDO, and I wish I could find a good one, but his correction value is for every 50pts above 150 take 2u of insulin (this is for meals). Well if a meter can be as much as 20 points off, that can real mess you up. So I've ignored his scale and did my own, and it seems to be working.

I understand the real key here is consistency of readings, but why does it seem that the docs seem to go off of what your readings are? Like my first week I sent my readings in and they were a little screwy because I was just learning to count carbs. Well my doc said to increase my correction by 2 more units per 50pts above 150 and take an extra 5u at night for my lantus. I told them no, and waited till the next week, which by then the original scale worked. I guess my real problem is the doctor not the meter :)

w5wjp
04-24-2008, 06:57 PM
Accuracy - how close you get to the true value

Precision - how many times you get the same value

For example, if a sample has 100 mg of glucose and you get a reading of 100, you are 100% accurate.

If you get a reading of 200 ten times in a row you are very inaccurate but very precise.

Mich
04-24-2008, 08:27 PM
Wow. We have a bunch of electronics people and mathematicians. I'm impressed. Excellent explanations of the whole meter experience. Thank you to all.

I also depend on consistency on my meter. I know what a reading of 140 means in terms of what I have to do to get it down to what I view as 90. I know how I feel at each level. That's all that matters for me. I view managing my diabetes as not an exact science, unfortunately. I just do the best I can. I agree with Scratch, the next advance should be a cure.

But in the meanwhile, I'll be thinking of all of the workings of a meter Redlan, every time I put a strip in.;)

Scrabblechick
04-24-2008, 08:32 PM
I'm like you, Mich! I think of everything that meter is doing as it's counting down! LOL.