The relative accuracy of blood glucose meters has of course been in the spotlight of late, since the FDA held a hearing to “crack down” on this issue last Fall. Funny, but I always assumed this had more to do with the technology inside the meter than the test strips themselves. Silly me.
One of this year’s buzzwords at the ADA Conference in Orlando last week was the “PQQ Enzyme” {glucose dehydrogenase (GDH)-pyrroloquino-linequinone (PQQ)} – a gremlin, it seems, that has 
been on the radar of meter manufacturers since 2003, when a medical device alert was issued in the UK warning that test strips containing the enzyme could could interfere with BG monitoring systems, causing false highs that might lead patients to overdose on insulin. (When test strips with PQQ are used, maltose or icodextrin — which is converted to maltose — galactose, and xylose are misinterpreted as glucose. Read the background on how it works here.)
Last summer, the FDA finally jumped on this, issuing an ultimatum making it clear that blood glucose monitoring companies were going to have to change the enzymes in their strips. Somewhat surprisingly, analyst Kelly Close writes in her report that the issue has “little impact to most patients.”
Now that meter accuracy has become such a burning topic, I might beg to disagree.
J&J jumped on this issue immediately, issuing an upgrade to its OneTouch test strips last fall. Now Abbott Diabetes has just introduced its newest FreeStyle Lite strips, which the company touting as PQQ-free. What’s also new on the strips are pointy little “ZipWik” tabs on each side of the strip, 
protruding from that dark half-circle where you’re supposed to apply the blood. These tabs are designed to make blood absorption easier, and as a longtime FreeStyle customer, I say, hallelujah! The fact that FreeStyle requires such a tiny blood sample is of course the reason I love them, but it often takes a lot of smearing and scraping and switching sides to get their current strips to recognize my blood drop (TMI? If you’re reading this blog, you gotta have the stomach for some blood talk, sorry)
Anyhow, surprised myself at how excited I was about the look of these new strips, with that pretty little FreeStyle butterfly right there on each one. And I’m glad know that they’ve done away with at least one of the major recognized causes of testing inaccuracy. I assure you the Abbott reps in the ADA booth were over the moon about this.
Overheard elsewhere at ADA: some folks from Bayer quietly bashing Roche for not being on top of the PQQ issue. When I reminded them that the Roche Accu-Chek folks had presented a white paper to the FDA recommending industry-wide steps toward improving accuracy, the response I got was:
“’Total system accuracy’ doesn’t mean anything if you haven’t accounted for the PQQ enzyme. We’re addressing it. I don’t know what they’re doing…”
{insert cat fight noises}
The upshot being: the PQQ enzyme is a hot-button that none of the manufacturers can afford to ignore.
Meanwhile, we patients are impatiently awaiting word from the FDA on how much tighter we can expect the acceptable margin of error to become. Current requirements are ± 20% error margin ninety-five percent of the time, which seems unacceptably loose, I’ll agree. But we must also understand that there are limitations in the science of measurement, and perfection is not an option.
Bringing this point home exactly was an activity I helped run at the Roche Social Media Summit, which took place the day after ADA concluded. (The activity was conceptualized by the Roche team – all I did was play MC). Using a fun “fact or fiction” format, we were divided into six small groups and tasked with deciding where accuracy is most important for us – at the low or high end of glucose measurement?
Now before you jump on this and say, ‘of course we want both!’ Remember what I just said about limitations in science: these systems currently can’t achieve 100% accuracy. The way I understand it, manufacturers essentially have a choice of where to shift the range they can achieve: either down, toward the lows, so the meter will be ultra-accurate at that end, but less precise on the high end, or vice-versa.
As reported already by David Mendosa and Sara Knicks: “the room was all over the map on this one!” That is, our group quickly realized how messy this issue is: the type 2 diabetics in the room were generally more concerned about recognizing highs, but some felt that a 190 vs. a 220 reading didn’t mean much since they’d address both the same way, while potential lows were scary to them.
The type 1s were really battling it out: instinct says we need killer accuracy for hypoglycemia, BUT false highs lead to insulin overdosing, which causes unnecessary lows. Not good.
Now It’s Your Turn: Assuming that the industry can conquer the “PQQ gremlin” and tighten up the range of BG meter accuracy, and assuming that Roche is correct in asserting that we need to “pick a range” to focus on, which of the following would you choose -
In the end, our group pretty much agreed that 15-15, the middle ground, makes the most sense. But you know what they say about compromise: nobody comes away happy.
Amy,
I would have to say that my gut reaction is that Roche wants to ride its current strip technology for as long as it can and that this is a significant driver in their “range” concept. I felt the same way when you posted the Roche “White Paper”, which is generally good (who can argue with better accuracy?), but has a “marketing-feel” to it.
In short then, Roche needs to address the PQQ “issue”.
FWIW, I would vote for 10/20, because that gives you greater accuracy at the low end while leaving the high end unchanged from what we currently have. My guess is that Roche can do any one of these steps way more inexpensively than redesigning their strips to reach 10/10.
Fair Winds,
Mike
Will the regular Freestyle strips that are used in the OmniPod PDM also have this new wicking tabs?
We’ve transitioned to using the OmniPod PDM to test blood sugars and use the regular, not Lite strips now.
If I’m reading the Abbott press release right, the FreeStyle strips (in addition to the FreeStyle Lite) will be compatible in all of their meters and also have the wicking tab and do not do not use the GDH-PQQ1 enzyme.
Am I reading that right?
@Leighann – my understanding is that the new ZipWik strips will be compatible with ALL devices that take FreeStyle strips. At least that’s what they told me on the show floor at ADA.
Wow, thanks for such a great write-up about this issue. The Roche activity you helped organize was really eye-opening and educational on how just twisty our test strips really are.
Thanks!
Ginger
An even better, but potentially more costly option would be test strips that fit your use case. Offer a product for those who are more concerned with high precision at the low end of the BG spectrum, one for the high end of the spectrum and one with greater precision at the “normal” range, but loses accuracy as it reaches the extremes.
This gives patients the flexibility to use strips to gain a better idea at all points.
why can’t they make two kinds — one more accurate on the highs, one on the lows? That way the patient/consumer can pick? Because everyone’s diabetes is different and every pwd has a right to chose their best treatment.
My feelings at first were the lows but when I realized (after some discussion) is that highs are where the long term damage is done. We need to get out of those highs as quickly as we can so knowing exactly where we are when we are in the 200′s and 300′s seems more of a concern.
I mean, I would treat a 48 the same way I treat a 62. 15g of carbs then test. But I would treat 220 different than I treat a 160.
I am really glad this situation is being addressed.
Recently before breakfast I tested at 235 (which didn’t make sense) so I tested again and it was 90!! Imagine what would have happened had I taken an insulin bolus for a 235 and my carbs. Definitely Urgent Care time. By the way, I did test two hours later to make sure the 90 was accurate.
I have gotten to the point where each time I have a high, I test twice.
Joan
good to know that the error rate is minimal, because this way we can improve a lot, good blog ….
emericle asked about 2 different types of strips. That is a good question, how would the accuracy, either high-end or low-end, be increased? Would it be thru strip technology, meter technology or both?
If it is thru the meter, could it have different modes? High-mode and low-mode?
How would increased accuracy at either end affect readings in the middle?
This is an interesting discussion of measurement accuracy over full glucose range which has been going on in one for or another for at least the last 25 years. All of the systems ever developed and marketed have had to deal with the trade-offs between user convenience, speed of system response, accuracy and reproducibility over full glucose measurement range, and measurement cost as it is impossible to simultaneously optimize all 4 factors at the same time. My own experience was as a chemist helping to develop the Direct 30/30 meter [marketed through Eli Lilly from 1989-1991]. This was a meter based on an electrochemical sensor which was reusable over the course of weeks and even months. Good analytical performance with comparable accuracy over a wide measurement range [40 to 400 mg/dl] and way ahead of its time in cost per test for high-frequency users since it was reusable, but with a slower measurement time relative to today’s strips [30 vs 5 sec], requiring a warm-up of a few hours to allow the sensor membrane to stably hydrate, and needing daily user maintenance to keep the sensor membrane area properly hydrated with daily calibration rechecks to verify continuing accuracy. This led to minimal market acceptance as users demanded total freedom from maintenance [comparable to the convenience delivered by a strip meter] as well as measurement speed and reasonable accuracy. Since this was pre-DCCT times, high- frequency testing was minimally practiced and the potential testing cost advantage was not considered so important. [FYI- These principles carry over in large part to the way CGM sensors work today, except of course the patient interstitial fluid acts to keep the sensor membrane constantly hydrated in a more consistent environment so less frequent recalibration is usually needed to verify accuracy].
But to some of the questions posed in this blog: Why the accuracy difference for sensor strips over different parts of the measurement spectrum, and how can two strips from the same packet show such large variability for essentially the same sample?
As a start, for cost reasons, most strip developers have optimized/minimized the quantity of enzyme reagents [reactants and cofactors] necessary to allow adequate signal development in the devices measurement time window. This usually implies optimizing for 75-250 mg/dl as this is where most readings will fall. When glucose is below these levels there may be little sensor signal relative to system background signal and depending on measurement data handling algorithms a high percentage error can result. The faster the measurement time the higher the probability for this error since sensor signal is less fully developed and makes up an even smaller part of the overall signal than if the signal was allowed to develop longer. At much higher glucose levels, the possibility also exists that one of the reagents will be exhausted prior to completing the measurement timing cycle so observed change in signal level is not representative of sample and the data treatment algorithm will be tricked to introduce an unpredictable error. Additionally, small artifacts in sample introduction to the strip can cause the device to trigger measurement of starting and ending signal levels at the wrong times so that the signal change measured is not representative of the true sample glucose. Again, the shorter the measurement time the larger the error from sampling artifacts can be. This can cause strange outliers in either direction across all glucose ranges and probably explains the variability described by Joan in one of the blog posts for essentially identical samples on two separate strips. Like it or not, the possibility of user technique artifacts exists and can occasionally invalidate the accuracy of a strip measurement. Of course, a great deal of work goes into the strip development to minimize the possibility of this happening but technique sensitivity can never be completely eliminated. Additionally, since the measurement consumes the strip reagents it is impossible to quantify the cause of the problem later on so both developer and user blame each other for the failure [i.e. The never-ending argument: Was it just an incorrectly manufactured strip or poor user technique? No way to tell].
As to the idea of multiple strips for different measurement ranges: certainly possible in principle but probably with insufficient market size to warrent the time, effort, and money needed for the development. We actually proposed a device to measure low glucose in pediatric patients 20 years ago but were told it was a trivial follow-up to previously developed technology and the market was too small to matter.
I found diabetic test strips WAY below pharmacy prices ($15/box!)
Hi all,
Thought this could help someone…I did a Google search for “cheap diabetic test strips” and came across this site that led me to a fellow diabetic that had my Freestyle Litr strips (50 ct) for $15 a box (plus S/H), and they don’t expire until the end of next year! What a godsend!
The site is http://www.webscheapestteststrips.com , and they also have One Touch, Contour, etc. Thanks
Alice
While accuracy is important, I suggest that it is a distraction from the discussion of COST.
Manufacturers would love to have all of us arguing about which meter technology is the most accurate because we aren’t talking any longer about how each test strip costs more than $1 apiece.
To put the issue of accuracy in context, spend a few test strip dollars by testing multiple fingers of each hand within a 30 second period — then ask yourself, “what’s the fuss about accuracy compared to cost?”
Hey; as one whose readinsg go 40 to 100 points off due to what I am eating and the man made sugars in the stuff; I object to this cavalier attitude about accuracy.
Accuracy that changes as I eat is crap and so are the strips that jump off.
Yes in am all meters read well but eat and wait 2 hours and see where one is.
This is not a discussion about accuracy but one of reliable consistent results. My liver does not play nice and shoots the non glucose D sugars into the main blood system for another pass thru the liver and past my fingertips to throw meters off scent that cannot meaure glucose d properly.
A meter that works consistently but has some reading issues but are consistent, I can live with and compensate. Something that is bumping around 40 to 100 points off based on the food I eat and the man made sugars stuffed in the stuff is totally unpredictable.
Please – lets have some rational discussion and get the slug strip technology out of the loop.