Sweet red blood

See this?

Measure 1 dl

This is a deciliter (about 3.3 fluid ounces) of “blood”. Where I live, we measure blood glucose concentrations based on how much glucose is present in this amount of blood.

(Actually, it’s not blood, it’s red wine. Humor me.)

So, if I wanted to find out how much glucose was in the above sample of “blood”, I’d simply extract the glucose, put it on a scale, and weigh it.

(Your nation may vary, and if it does, I apologize in advance for not doing the conversion for you. But mmols involve counting molecules, not weighing them, and mmol folks base it on a full liter – or litre if you prefer – which is an order of magnitude larger)

104 milligrams of glucose in that deciliter of  blood? Perfect. 104 mg/dl.

Except nobody in their right mind would want to draw THAT much blood – perhaps ten times a day – to keep tabs on things. So what do we do? We sample – we take a smaller amount and extrapolate the results.

For instance, instead of a full deciliter, we could take one one-hundredth of that amount of blood, extract and weigh the glucose, and take the result … times 100.

Measure 1 ml

One milliliter of “blood”, circa 2010

Of course, if our measurement is off by a little bit, that error gets magnified by a lot when you do the math.

But it’s still a lot of blood. So why not take one one-hundredth of THAT amount?

To get your blood glucose level, simply draw this amount of blood and multiply the weight by 10,000. If your weight is off by just 0.01 milligrams, your final result is off by 100 mg/dl. Unacceptable.

Measure 1u

One hundredth of a millilter of blood, 13.5% alcohol by volume, circa 2010

But even that blood sample is unacceptable.

According to this recent Diabetes Forecast comparison (pdf), some of the most popular meters use samples in the neighborhood of 0.3 microliters (Abbott Freestyle) to 1.0 microliters (OneTouch Ultra)

If I’ve lost you with the math, let me summarize: there are one hundred-thousand microliters in a deciliter.

This is one microliter, the amount of blood needed for a OneTouch Ultra meter. It’s mighty tiny – you might be able to detect it in the second photo.

Measure point-1u pump

Do not, UNDER ANY CIRCUMSTANCE, infuse red wine into your body subcutaneously

Measure point-1u infset

or supercutaneously (whatever that means)

You get the idea. Weigh the glucose, multiply by one-hundred thousand. A tiny error can make a big difference.

Now, of course, this isn’t how we really get our blood sugar readings.  Not at home anyway (Maybe at a lab, I really have no idea). Modern meters mix the blood with a chemical, zap it with electricity, crunch it through a mathematical machine, and out comes mg/dl (accurate to 20 percent or so). To be honest, considering the size of the sample, they do it remarkably well.

Yet we’re still not happy, and justifiably so.  Well enough is not good enough.

Accuracy matters.

But is it realistic? Can we really get accurate blood sugar readings from such a small sample-size?

“But blood is not fixed. It’s liquid. It’s dynamic. […] Your heart pumps it both upstream and downstream. […] For our entire diabetic careers, we have been indoctrinated, trained, brainwashed, and bamboozled by our only available technology, the fingerstick meter, to view an inherently fluid part of our body as something static and fixed.”

-Wil Dubois, Beyond Fingersticks, Chapter 5

In other words, the glucose concentration at your fingertip is not the same as it is in your carotid artery, or your superior vena cava, or your forearm. (But you knew about the fingertip/forearm discrepancy. It’s in every standard disclaimer about alternate site testing).

I suppose it could even vary from one drop to the next. One microliter to the next.

The smaller the sample, the more room for variation.

I fully support the demand for more accurate glucose testing. I encourage you to, too. It’s important, and there’s a call-to-action going on RIGHT NOW!!!

But in the back of my mind, I still wonder — at what cost do we demand it?

Will accuracy-driven regulations mean we need to go back to a sample size of:

This?  Measure 1u

Or this?  Measure 1 ml

Or this?  Measure 1 dl

* * *

PS – That insulin pump I used above is an old, long-out-of-warranty pump that hasn’t felt the charge of a battery in years. The reservoir and tubing are the byproducts of a set-change a couple of days ago. And the syringe is about a decade old, purchased before I started using insulin pens and before I started pumping. I’ve got a ton of ’em lying around. My own diabetes care (however unbalanced it has been as of late, though I understand there’s a book for that, which I fully intend to get my hands on) was completely unaffected in the writing of this post.

PPS – The wine was from a surprise 40th birthday party my wife threw for me last weekend (I’m really old). My own diabetes care was affected by that.

PPPS- Speaking of books, that edited paragraph I excerpted from Beyond Fingersticks: The Art of Continuous Glucose Monitoring came from a paragraph about expectations, and it totally changed the way I look at not only CGM, but blood sugars in general. If you haven’t read it, I highly recommend it. Brought to you by the author of Ask D’Mine.

PPPPS – Someone once said to be careful what you wish for, ’cause you just might get it. So when I make my demands known (do it here, do it now!), I’m not only asking for better accuracy, but no increase in sample size.

Posted on March 10, 2014, in Diabetes and tagged , , , . Bookmark the permalink. 11 Comments.

  1. Wow that’s pretty eye opening! Thanks for sharing this. And happy birthday!


  2. Totally awesome explanation, Scott! I guess I’m one of the very few that still draw out a pretty large sample when I test… not intentionally, but it’s somewhere in my method or maybe I see the small sample and I am still of the old mindset that “whoa! that’s not enough! SQUEEEEEEEEEEZEEEEEE!!” or the old “milk the cow before pricking” and I end up with a TON of it. But I certainly wouldn’t want them to change the sample size because I know that wouldn’t be good for the other 99.999999999% of people who like the smaller size that’s required, like d-parents (and rightfully so).

    Liked by 1 person

    • No, not at all. I still get a larger-than-needed sample as well. That way, it’s easier to line up the drop with the edge of the test strip (perhaps the 1 microliter – or less – is what gets sucked into the strip, but it’s quite difficult to make it work with just that amount). But I admit that I was in disbelief when I saw how much 1 microliter (0.1 u) was… I did it about five times to be sure it was right.


  3. Happy Birthday

    Thanks for the StripSafely links.

    I love two comments you make,
    “… they do it remarkably well.” Yes most do and there should be no place for this that don’t.

    “…at what cost do we demand it?” It being better performance. Excellent question.

    I think these two are deeply related and the whole point. The StripSafely effort calling for enforcement opposes sacrificing ‘remarkably well’ for cheap. It is unacceptable to trade off the real world performance of devices for pennies. A price only view drives the degradation of quality that the market is seeing.

    I agree, it is also incumbent upon users to appreciate the trade cost benefit relationship going forward. Accuracy can and should improve in an economically responsible fashion. The investment in R&D should realistically returned to investors. I see incremental progress as the means to do that. I think the FDA’s proposal for personal use offers incremental advances that are realistic and can be economically sustained. I worry that the in clinic/hospital requirement proposed by FDA may not be either possible or economical, based on comments by prominent experts.

    Key to incrementally and affordably better is enforcement that keeps the devices that fail to perform to standards out of the marketplace and mandating that the very dangerous drug, insulin be dosed with only the incrementally better devices.

    You point out how small a unit if insulin is, too much is literally just a drop. The calculus of cost should consider the high price of insulin overdose including the cost of spent for ER admissions and more. The 2003 standards should not move forward past 2014, technology can and should advance. As it advances the conversation should look froward and value the benefit of better accuracy.

    The real value of success can be measured in the manny happy returns of the birthdays of millions of people made possible by doing incrementally more remarkably well managing diabetes with insulin.

    Manny Happy Returns.


    • Thank you Bennet.

      By “at what cost”, I was actually referring not to the monetary cost or the research investment, but the cost in blood – literally. When a mandate comes down from the federal government, those affected tend to look for the easiest way to comply. Hence, the reason new requirements with regards to schools, highways, and trains tend to be met with increases in taxes, tolls, and fares. (In my professional life, I tend to work a lot with these so-called authorities which are constantly barraged with unfunded mandates).

      When a company must comply with new accuracy requirements or get pulled from the market, the easiest way I can see to meet those requirements it to increase the sample size. Sure, they can do more research and improve quality-control, but the bigger sample size would require the least research and would offer more likelihood of success.

      I just don’t want the meter manufacturers to take the easy way out, that’s all. If I can get 20% accuracy with 0.5-microliter sample or 10% accuracy with a 1.5-microliter sample, it would be tough to convince me to go with the latter.


  4. The best part of this, other than the great message and education….are all the PPPPPPPSSS’s.

    I love them. Clever 🙂


  5. Scott – well said.

    I have been a T1 for 30 years. I also have a degree in Chemistry, so the concept of chemical measurement is something I understand inside and out.

    Working in the environmental testing world for many years – 20% relative percent difference (RPD) for duplicate analyses is considered acceptable and proof that your methods are working within an acceptable range. Every measurement has sources of error – your examples show many of the ways a measurement can go wrong.

    Obviously – when our health and lives sometimes depend upon these numbers – we would like them to be perfect representations of what is going on inside our bodies. This is truly not possible, based on the reasons you mention. But we would like a better RPD (ie better than 64 to 96 when the real number is 80). We would like to approach perfection.

    As you approach perfection, the problem is going to be the cost involved. Analytical equipment can be made more accurate. it can be made faster. It can be made to use a smaller sample size. The best we can hope for is the sweet spot where each of these variables is acceptable and then the technology is usable by a child without a degree in Chemistry. This is incredible hard to do.

    I think people have come to expect our glucometers to work like Star Trek type tricorders. That just is not possible. We can strive for this goal – but we do need to be realistic, which is what I think you are talking about.

    Liked by 1 person

    • Thanks for the reply Brian. Like you said, there are multiple factors at work here — one is the technology to measure with such precision, but another is how the blood sugar concentration simply isn’t uniform throughout the body. If the real honest-to-goodness-100%-accurate blood sugar is 80 in the left index finger, but is 76 on the pinkie or 85 on the right thumb, then what? No improvement in measurement precision can address that; you just need more blood from more places.


  6. Happy birthday!

    I’ve never understood or even thought about the whole dL thing. HUGE amount of blood. Why not just call it mg/L and normal could be 1000?


  1. Pingback: Awkwardly avoiding awkwardness | Rolling in the D


Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s