CGM: Waiting for Godot

More thoughts on continuous glucose monitoring (CGM) here: Like you, I anxiously await the mainstreaming of this utopia of diabetes management.

It’s great to see progress, like European approval of Abbott’s much-hailed Navigator system, and FDA approval of a seven-day-wear version for DexCom. (The company will host an open conference call to discuss the new product on Monday, June 11, at 1:30pm Pacific.)

As noted yesterday, however, I believe the benefits of CGM aren’t quite outweighing the drawbacks for users just yet.

Dr. Martha Nolte of UC San Francisco summed it all up nicely in a recent presentation at that university’s annual Diabetes Patient Symposium:

CGMS – Does it help?” one of her slides asked, rhetorically, and then:


• Alarms for high and low blood sugars and
trend alarms allow earlier intervention,
prevent more severe highs and lows.

• 24-hour trend data helps fine tune insulin
dosing algorithms

• Improves HbA1c

All good stuff. BUT… followed by a number of slides outlining the drawbacks, including inconvenience, low accuracy and high cost.

For specifics, check out these comparison charts* she presented:



(* Don’t ask me why Dr. Nolte included the GlucoWatch at all. I thought that product was washed up long ago due to problems with burning patients’ skin.)

Start-up time on the newer CGMs is two to ten hours, and the best accuracy for an FDA approved model is just 72%, which means nearly 30% innaccuracy. Ugh.

In addition, Dr. Nolte says “all of the CGMS devices miss low glucoses.” Which sucks, considering that detecting lows is (almost) the whole reason we’re so hot to use the systems in the first place.

As you all know, I tried the original DexCom for five months and found it to be a great idea whose time hasn’t quite come. It requires four or more fingerstick calibrations per day, which are possible using only one specific OneTouch meter model (which I happened to hate — required too much blood and wasted a lot of test strips). My unit also has extremely loud alarms that cannot be adjusted or disabled, and since the accuracy isn’t all that great, the unit would often beep my husband and I awake all night for no good reason. I mostly keep it in my closet now, awaiting obsolence by Version 2.0. CGM made liveable…

Oh, how we pin our faith in a better diabetic existence on this nebulous force. But we are justified in our high hopes, I believe.


10 Responses

  1. Ed
    Ed June 7, 2007 at 7:26 am | | Reply

    My biggest excitement for CGMS is for athletics. I really don’t mind doing the finger pricks 10 times a day but when you’re active its tough to do. I coach in a youth football league and have been worried on several occasions that I’m going low during practice – rather than testing myself in front of 15 10 year olds I just choose to take a bite of a cliff bar – it would be nice to know that I am going low and not just getting a headache out of yelling too much!

    Also, CGMS would be awesome for runs and bike rides, where I don’t need 100% accuracy but do need a trend analysis to know when to pop an energy Gu or have a sip of gatorade. Of course I’d love not to prick my fingers anymore but some constant guidance would be good enough for me at this point.

  2. Kassie
    Kassie June 7, 2007 at 7:59 am | | Reply

    Amy – I’ve seen you mention a few times (here and in previous posts) that the dexcom requires 4 finger sticks/day. Mine (which I got in January) only requires one every 12 hours.

    Also, I don’t think that low detection is the biggest benefit/appeal. I know that it wasn’t for me – I was more interested in avoiding extended overnight highs and in the ‘policing’ aspect of the device. Accuracy affects both of those, of course.

  3. Bernard Farrell
    Bernard Farrell June 7, 2007 at 11:54 am | | Reply


    We’ve two new ones on the horizon. The Dexcom SEVEN and the Abbott Freestyle. I think these will both be big improvements, but not huge leaps forward.

    The question I ask myself is. Should I use this while waiting for something better to come along? And for the Dexcom (which has definitely not been smooth sailing for me), the answer is yes.

    I’d like to be still alive when the ‘cure’ is found. So in the mean time I’ll work to maintain a semblance of control.

  4. Sarah
    Sarah June 7, 2007 at 12:01 pm | | Reply

    While I think that CGM’s are a step forward for the most part, reading this also confirms what I think…We are *nowhere* near an “artificial pancreas” (at least one that won’t kill us), no matter what JDRF claims when they solicit you for money.

    In fact, I think diabetes will be “cured” by other means before a true closed loop system ever is successful. In fact, one Medtronic rep agreed when I said that there will *never* be a truly closed loop system…it’s too much of a liability. We all know how machines can fail. One mistake and a diabetic’s car has killed 5 Medtronic is responsible for 5 deaths.

    CGM’s are great, but they are not a cure nor a perm. solution. They are simply another expensive device that we as diabetics must pay for monthly in terms of supplies until we die. Trust me, if diabetes technology is so far behind today (we can build a space station but can’t perfect a CGM or non-invasive glucose testing), I have little hope that any of this will be much better in our lifetimes.

    Let’s stop feeding the pockets of Big Pharma and demand a real cure, not technology to make diabetes more “controllable”.

  5. Brian
    Brian June 7, 2007 at 12:06 pm | | Reply

    Amy, you mention that the drawbacks are too great for you at the current time. Could you explain what improvements (cost, performance, ease of use) you would require before you would consider using one again? Personally, the old real drawback I have currently is with the cost. Comparing the functionality to glucose meters is similar to comparing glucose meters to urine test strips in my opinion! I imagine everyone has their preconceived threshhold in making the jump and I would simply be interested in yours.

  6. AmyT
    AmyT June 7, 2007 at 6:19 pm | | Reply

    Brian — that would be cost, performance, ease of use. Actually in reverse order. Thanks for asking.

  7. Jen
    Jen June 7, 2007 at 9:17 pm | | Reply

    I have the MiniMed CGM to go with my pump. I loved the fact that I didn’t have to test as frequently (I was testing up to 12 times a day before) and I could literally see how food and exercise effected my blood sugar, which has been really helpful. But, oh my god, I agree Amy, the advantages don’t outweigh the disadvantages. Mine was painful to insert, painful to have in, and I was so taped up with the monitor part that I couldn’t hardly move. I’m not sure if anyone else had the older MiniMed one, but the monitor is huge. Also, $1000. And then the sensors which are $350 a month (if you abide by the 3 day only rule). Now it seems a waste in that I haven’t used it in months…

    Someone is definitely going to have to make vast improvements before I spend my money on one again, but I certainly will buy one when they’re easier to use.

  8. Jack Sallen
    Jack Sallen June 8, 2007 at 10:19 am | | Reply


    I have done a good deal of research into glucose sensor technology. Unfortunately, it appears no one has it right. The use of interstitial fluid is one of the best examples of such research that will not result in any usable solution regardless of the amount of money sunk into such projects as it suffers not only from latent readings of up to thirty minutes but also requires recalibration several times a day. Clearly, such a system would never replace test strips for instant, accurate readings which is really the point of all these endeavors.

    As I am sure you are aware, most glucose tests today measure blood resistance. Other factors such as hydration and salt levels can distort such tests. However, new methods of measuring glucose that do not rely solely on the resistance of blood are, at least, promising alternatives.

    Several such sensors exist including the micro/nano fabricated glucose sensor found at After discussing this proposal with a PhD in physics, I learned this solution also fails to deliver in its promises. The fundamental issue appears to be the need to “flush” any remaining hydrogen atoms that remain in the nano tube. Such “flushing” is quite tricky as the hydrogen has a tendency to not want to leave the tube.

    A different take on continuous glucose monitoring can be found at . Essentially this sensor is implanted in the body along with a micro magnetic coil. The magnetic coil is designed to vibrate at a frequency correlated to one’s glucose level. Alas, this too has a fundamental issue regarding fibrosis and one’s body rejecting the device as a foreign substance. Although anti-rejection medication, due to its intolerable effects on the body is obviously not the solution to such a problem, artificial heart research has yielded working solutions to such issues by finding certain polymer coatings that essentially hide the device from one’s immune system.

    My take on the situation is that much of the research is hopelessly flawed and most of it is headed down a path whose stated goals fall short of current glucose monitoring sensors in latency, accuracy and even price. All this, when one takes into account the fact that current sensors are themselves overpriced, is truly disheartening.

    The state of current and future glucose sensors I fear are near criminal. One can not be blamed for even questionioning the very motives of the bio industry after dealing with or using their “solutions”. As a user of the accu-chek compact, manufactured by Roche Diagnostics, I was interested in retrieving the stored readings via the infrared port on the device. The software, already being quite unusable, had a price of $25. Deciding I could at least write my own recording and graphing software and, even if to a small degree, gain a grain of control over my glucose monitoring and perhaps. I contacted Roche Diagnostics who outright refused to even provide a specification for the protocol used between the monitor and the computer. The attitude towards the customers of these manufacturers befit that of a French monarch.

    Jack Sallen

  9. Titos
    Titos June 8, 2007 at 11:58 am | | Reply

    I think people are being too pessimistic. This is first generation technology and improvements have been major and rapid, both in quality control as well as performance (e.g. the minimed cgms for kids or the dexcom 7). The difference with the past is that it seems that the underlying technical challenges of cgms have been cracked and things like automatic calibration or accuracy will be solved over the next months (see dexcom announcements on a new product for patients in intensive care). However broad usage of cgms will come after it has been tried and tested in “critical applications” where tight control and low avoidance are very important. Examples:
    - pregnancy
    - intensive care
    - sports and tournaments
    - driving or working with machines
    - exam periods
    In the above cases the inconveniences of a cgms are relatively unimportant compared to the benefits and trends are more importnat than absolute BG values. These applications will provide the companies the info they need to improve the products. CGMS will have wide spread use within 3-5 years – more than pumps.
    On the artificial pancreas, it is highly improbable that a practical closed loop system will be developed anytime soon, ie with a sensor and a pump. For the simple reason that it takes insulin too long to have an effect and the sensor can only detect BG after the effect – unlike the brain/pancreas. However there are other ways of developing the equivalent of an artificial pancreas, e.g. by developing insulins that become effective according to the glucose level and transmit the BG information from inside the body, in essence putting the intelligence into the drug itself. It is encouraging that these developments are already underway.

  10. Titos
    Titos June 8, 2007 at 12:32 pm | | Reply

    I’ve heard this problem of interstitial fluid mentioned so many times by so many people…
    the point is the variance is highly stable and can be predicted statistically. That is why the accuracy of cgms will improve -they will possibly be more accurate – because of the frequency of the readings – than single fingersticks when compared to lab readings. And in any case what do you prefer – highly accurate fingesrticks 10x / day or slightly less accurate cgms readings 500-1000 x / day and trends if you want tight control.
    There are many non-invasive or minimally invasive technologies like the ones you describe on the drawing board and a lot of challenges still to be met. They will be met once the current technolgy proves there is a market out there for the substantial investments required to make economic sense

Leave a Reply