More details on the recent FDA Hearings on the accuracy of blood glucose monitoring devices: I was surprised by the report from patient advocate Ellen Ullman on Monday, noting that patient interests seem to be grossly underrepresented, and lots of experts seem to believe that the status quo of +/- 20% error margin is all that any of us need (or can handle). Grrrrr….
See Dear FDA over at “Your Diabetes May Vary” for an earful of why the status quo is NOT OK.
OMG - Who's seen this appalling ad?! Do not get me started.
Meanwhile, I was lucky to be privy to a White Paper submitted to the FDA on the topic by Accu-Check/Roche Diabetes Care. (Yes, the folks who host the Diabetes Social Media Summit, but no – this can’t be just for publicity’s sake, since the paper is not publicly published.)
In this paper, Roche makes some very smart and reasonable recommendations in six areas — things that kind of shocked me, only in the sense that they’re not in place in this industry yet:
1. Accuracy
“The standard for allowable deviation in blood glucose readings — as set by the International Standard Organization — should be strengthened in two important ways:
- by changing the breakpoint (mathematical value used to calculate averages) from 75 mg/dL to 100 mg/dL to promote a ‘balanced bias’ for BG values at both the high and low-end
- and by reducing the current allowable variation down to +/-15%
Of course, we patients are pushing for +/- 10% ourselves!
2. Interferences and System Limitation
“All manufacturers should be required to test for at least the 100 most common interferences” (such as skin oil, food residue, air temperature, etc.) that can result in faulty BG readings.
No standard currently exists for this! Also, per the paper: “The manner in which interferences are tested should be standardized.” Seems like a no-brainer, ay?
3. Safeguards
Glucose meters should be required to include “failsafes for common incidents, such as not enough blood on the strip, or expired or damaged test strips.”
The way I understand this one, it means an actual explanation on the screen rather than the cryptic ‘Error 1′ or ‘Error 3′ we get now.
4. Labeling
“Consistent, user-friendly standards regarding product labels — similar to those found on food items, from bread to soda to cereal — should be adopted.”
This has got to be an improvement over those tiny-print package inserts we get now, which are expensive to produce and promptly discarded without being read by most of us.
5. Quality Assurance
Specific guidelines for how user complaints and ‘reportable adverse events’ for glucose meters should be handled.
No guidelines currently exist on what even constitutes a reportable event, or on manufacturing processes for blood glucose systems, according to this paper. For such a high-value industry, Yipes! is all that comes to mind…
6. Support and Education
“Standards for comprehensive product support and education should be available to support self-testing… Currently, there are no mandates for either.”
I believe the idea is to go beyond simply fielding customer phone calls about product failures, to actually teach people how to use their meters to impact their health. By golly, we need this!!
These all sound like fantastic improvements to me. Again, I can hardly get my head around the fact that this stuff hasn’t been addressed already.
As my D-blogger buddy Bennet notes, glucose meters are powerful tools to help us stay healthy. What we need now are accuracy standards that envision our daily lives, meet the actual use of our meters and progress with technology (!)
WoHoo I’m D-blogger buddy and that sound like a Colbert ‘Friend of the Show’ to me. LOL
Thanks for keeping this in the spotlight Amy.
Please everyone, drop the FDA a little note.
Amy,
Call me a cynic, but as good as the Accu-Chek white paper sounds, it might also just fit the way they are currently doing things and/or planning to go. Also, it could raise ‘barriers to entry’ by innovators (like those submitting on your awesome “2010 Design Challenge”).
Now my little pet peeve on “expiration dates”. The Accu-Chek dates are hard coded. No tests beyond the date on the vial, period. Wouldn’t it be nice instead to have the meter able to determine if the strip is viable?
Fair Winds,
Mike
If my BG were at 53, I wouldn’t be able to smile like that woman. But I get the point. Until the standards are raised, though, much of testing involves Dead Reckoning, for me. That means that I might base my trust on a reading, based on what the last and before last readings were…similar to sea navigation before the GPS was invented.
My pet peeve is that the instruments are being made with increasingly cheap materials, and the prices are going up, and accuracy fluctuates from stick to stick…sometimes by 50 units–hey, what’s going on here?
I’m a recent Diabetic, I had no idea that the glucose monitors had such a wide error margin. This scares me especially since I’m trying to get my insulin under control.
Let the insulin-ignorant T2 ask some baby questions here.
I get the impression from some of the T1s and other insulin-users that a 1 unit correction should lower blood glucose about 50mg/dl? (I imagine this depends on the person; I’m just doing some “ballpark math”.) And similarly, I get the impression that unless you’re on a pump, corrections of smaller than 1 unit, maybe 1/2 unit, are not easily administered?
Those two premises would suggest that for reasonable correction calculation, there should never be more than 25 mg/dl absolute error in a meter — even if the correct reading is 490 mg/dl. That translates to +/- 5% error at 500 mg/dl. The same percentage error at 100 mg/dl would be 5 mg/dl, which is a more comfortable margin of error for those of us who prefer to keep tight control (range 80-120 mg/dl), regardless of whether or not we use exogenous insulin.
I don’t know enough about how the human body responds to hypoglycemia to ask the right questions about the low end of the scale — but again, any margin greater than 5-10 points, especially at the hypoglycemic/euglycemic border, can result in an incorrect therapeutic response.
Yeah, no one mentioned that guy shouldn’t be wielding a knife at 53. Geez.
I can’t read the small print, but is it 53 mg/dL or could it possibly be 5.3 mmol?
It’s absolutely 5.3 mmol. I zoomed in using my browser.
I think the ad is risky and eye catching. It doesn’t bother me. The fact is, with 20% within the acutal the acceptable range, having the meter read wrong could be just as life threatening as having knives thrown at you.
Ok, slight exaggeration. I just don’t understand why there needs to be an uproar over this ad. It’s pretty and it makes it’s point. That’s advertising, no? Plus, it’s getting attention, another plus.
Hiya
It seems to be a Canadian ad (judging by the .ca on the website) which makes it 5.3mmol/L which translates to around 96mg/dl. Not quite diabetes perfection, so I wouldn’t be encouraging him to throw it either!
Good job Amy on raising awareness of an incredible issue. Can’t believe there are no standards on these seemingly basic aspects to testing that affect every single person (with d) and every single meter..
Particularly when there are so many guidelines are based on the number and treated as gospel: do not exercise if over 16mmol/L (288mg/dl), if ambo’s visit you at home for an errr middle of the night seizure, they won’t take you into hospital if you test over 4mmol/L (72), etc etc
Monique Hanley
Your article was most tweeted by Diabetes experts in the Twitterverse…
Come see other top popular articles surfaced by Diabetes experts!…
I have a lot of concerns with these changes in accuracy:
Will test strips soon be $5 a piece instead of the already outrageous $1 a piece?
Will those $5 test strips fail more often, requiring more frequent sticks as well as the extra expense?
What about the price of meters?
Will more companies just simply get out of the meter business because the standards are too high? (further driving up prices)
Will meters become more difficult to use? (“We are sorry. You must insert a new lancet before meter can be operated…”).
You see accuracy. I see more control being taken away from me about how I use my meter.
Hearing from “experts” that a 20% error margin is just fine for us non-experts is only demonstrating that said experts have a knowledge gap regarding patients, and have heard too much propaganda from manufacturers. I am not sure if there is any sort of medical analysis equipment currently in use where an error margin as large as 20% is considered permissible. What the 20% error margin allows is a considerable reduction in cost of both the meter and the test strips, as compared to some more accurate technology that would cost more to produce. The argument that manufacturers don’t want to spend R&D funds and then raise the production costs on a very profitable product makes considerably more sense to me than the explanation that users are either not interested or don’t need any more than 20% accuracy.
HemoCue makes a smallish (about the size of a paperback) meter that works similarly to typical BG meters. However, it achieves the same accuracy as the most sophisticated lab analysis equipment. It’s expensive and requires a 5 uL sample, but perhaps it’s possible to develop something that relaxes the accuracy to 5%, uses a smaller sample size, and has a lower production cost. As a matter of fact, David Mendosa reported in early 2006 that HemoCue was about to file a 510(k) on a consumer version of their meter. It had the same lab-standard accuracy as their larger meter, and was expected to sell for about $70.
http://www.diabeteshealth.com/read/2006/02/01/4515/hemocue-is-coming/
I have always wondered what happened to this product. It does show that 4 years ago, one manufacturer found it feasible to create a far more accurate meter aimed at the high end of the consumer market. Given advances in technology from 4 years ago, I can’t believe that a meter like this is not an achievable goal.
[...] to post below: Amy Tenderich (Diabetes Mine) has posted a couple updates on this issue, including a white paper released by Roche and an interview with Ellen Ullman, mother to a son with type 1 and the only patient advocate at [...]
Regarding recommendation #3 around safeguards- In the Ultra’s operating manual it says that if you don’t have enough blood you could get “an error message or a false reading”! Yikes. I think I’d rather have an error message!
Does anyone have a clue as to what it would take to get from 20% to 10% accuracy? Is there a technical limitation of the strip? Meter? Both?
Do we need more blood? Longer wait time? I’d easily cough up much more blood (and money) for +/- 5-10% accuracy.
I too am amazed that there are so many variables in what we have been trained to think of as accurate.
Is it any wonder why we go high or low when the math we’re using could be off so much? And this stuff doesn’t even address the trillion other variables we deal with.