A Technician's Suggestions On Using Dexcom

I have a friend named Hayley. She is a Type 1 and is studying to be an endocrinologist. She is also a technician. The following suggestions were posted by her on another site. She gave me permission to copy and past them here.

"Unfortunately there is not a lot of science to many of the recommendations for calibrating or use. The recommendations provided by the Dexcom recommendations stem from the approaches taken in studies used to obtain FDA approval for the device.

I have quite a bit of experience with this as I worked as a technical support representative and taught other physicians about the system for 18 months following graduation.

Here are some things that I have learned from my own experience & can be really helpful - AGAIN, this information is based on my experience/opionions and do not necessarily reflect those of Dexcom (liability disclaimer):

1. The sensor is waterproof but many times the adhesive is the length-limiting factor. The STS shower covers can preserve the adhesive and sensors can last for 2-3 weeks if in a "sweet spot.“

2. The system is only FDA-approved for the abdomen and so Dexcom reps are not supposed to give you information about inserting it anywhere else. However, this does not mean that it is necessarily most effective when placed in the abdomen - again, it was the only site studied.

3. Qualities of an ideal site include: adequate amount of subcutaneous tissue (fat), no scar tissue (like fibrous surgical scars), and ideally on the same side of your body as the receiver

4. The dexcom and sensor communicate using radio waves - not a very strong signal (and for good reason - we don’t want any sort of radiation going through us). For this reason, the receiver and transmitter do not communicate through your body. You can have a sensor on the left and receiver on the right and provided they are both on your front or back, they’ll communicate find. However, if you have a sensor on the front and receiver in your back pocket, you may intermittently receiver the Y symbol (for no signal)

5. The Dexcom receiver cannot calibrate when there are ??? in the status box. I think the algorithm makes it think there is no number so it can’t compare and effect change when a number is used. It doesn’t know if 124 is below or above ??? if that makes sense. It will also not calibrate with numbers <40 or >400 mg/dl

6. Some people swear by calibrating it frequently. However, more frequent calibrations do not necessarily make a huge difference in the accuracy. This is one fact that I cannot confirm or deny because I’ve never really closely examined it but the engineers state that 2 calibrations a day are all it needs, no more (unless prompted) and no less.

7. The reason that two calibrations are advised initially to give the Dexcom information about how your glucometer interprets the blood glucose at that time. It uses this with an algorithm to equate what the sensor is seeing to the blood glucose at that time. I don’t, nor have I EVER, been a big supporter of taking two fingersticks from two different hands (ouch!). The point of it is just to get a basis for what your glucometer interprets. If you were to test repeatedly within 1 minute, the glucometer will read the glucose level differently each time. Some glucometers are more consistent than others but there have been several studies published recently that examine the accuracy of the glucometers and there is definitely a margin of error.

8. Though not FDA-approved for use in children, it is a great device for kids because the sensors last longer, the sensors are smaller, the lag-time is 5 minutes (as opposed to up to 20-30 minutes for Minimed), and the blood glucose level does not have to be stable (unchanging) to calibrate. The one caveat with all of the CGMS products is that kids have much less fat so finding a good site is more difficult! However, a lot of parents use the buttocks.

9. Sensor insertion hurts (I think). Though going faster doesn’t necessarily ease this pain, it can increase the likelihood that the sensor will function. The longer it takes to go in, the more opportunity to get “caught” on parts of your skin.

10. if your sensor fails before day 7, has greater than 30% discrepancy with your glucometer or is consistently ??? before day 7, call Dexcom for a replacement. They are expensive little buggars!

11. Finally - if 2 sensors in a box fail, it does not mean the whole box is bad. Each sensor is made individually and the success of one sensor is independent of the others. If you have sequential sensor failures from the same box - there could be a problem with insertion or a problem with your transmitter (as transmitters get older, I’ve found that they start to have more frequent failures. This might be from the transmitter battery.)”

Thanks for the info, Richard!

This is really great info. I love hearing this kind of advice from someone who personally uses the dex, is studying to help others with diabetes, and has worked from a technical side of the dex, so knows what she is talking about. It makes me trust the advice much more. Thanks!

Thanks Richard,

Lot’s of good info!

I have a question about point number 8 if you have a chance to get clarification.

I know the Dexcom receiver takes a reading from the transmitter every 5 minutes. When she says “the lag-time is 5 minutes (as opposed to up to 20-30 minutes for the Minimed)”, is that what she means? Or, is she saying that the lag between the BG reading from whole blood and the interstitial fluid reading from the Dexcom is 5 minutes?



This is great information to share. Thank you and thanks to your Dr. friend for allowing us Dex users to gain more knowledge.

Love what is said at point 7, as it matches exactly my own experience of 16 months with the Dexcom Seven Plus.

I always insert twice the same number when the 2 hours of the first calibration lapse, and it’s always been working really well for me.

There was a discussion about this not long ago, and I am happy to see that others have came to my same conclusion regarding the 2 different (or same) readings to input into the receiver at sensor startup.

Ciao, Luca

FHS, my interpretation of Hayley’s #7 is that the Dex reading from the interstitial fluid is 5 minutes out of step with the current BG reading, as you suggested.

If any of you have questions you want me to ask Hayley, let me know. She does not post very often on the other site. She is a very busy lady. I could collect a few questions here and send her a private message there. I am sure she would get back to me eventually.

Thank you Richard, this is incredible information!! And it’s REALLY nice to know a fellow Type 1 is going to be an endo! We really need more of us becoming endo’s!! This information cleared up some grey areas I’ve been having since I started on the Dex! You have to let Hayley her information was MOST helpful!!

Thank you Richard!

Richard, I agree with most of these items. But I have some comments, too.

Regarding #1: There are much better, much cheaper protectors than Dexcom’s old “shower covers”. They’re widely discussed.

Regarding #4: Right! But also note: Going through your body side-to-side, or front-to-back, is not as likely to cause “Lost transmitter” than forcing the signal to be heard along a diagonal path. When people put receivers under their pillows, or on nightstands, the thickness of straight-line body tissue can be much longer.

Regarding #6: The CEO has already mentioned that changes will be made in the calibration schedule. (The current BiD schedule isn’t “optimal”, and can be improved very easily.) Support engineers who say that the calibrations should occur exactly per instructions (from the Dex) can’t discuss better calibration schedules – it’s EXACTLY like their inability to discuss “alternative sites”. I think that the first day or two need more frequent tests, and days 4-10 can get by with much less (as little as one per day). The Navigator’s schedule might be close to Dexcom’s next-generation scheme. YMMV, of course.

Regarding #8: The FDA has been concerned with usage in kids, and perhaps and not merely because it’s “off-label”. The wire breakage problem could be more frequent, or lead to more difficult complications. So let me add a bit of advice to avoid wire breakage: Always remove the Sensor and Dex adhesive tape by lifting the end with Transmitter clip “free arms” first. Here’s why: when you inserted the Sensor, the wire went straight in at an angle, and then got bent flat by inserting the Transmitter. Lifting from this side first reduces the angle and straightens the wire, making it less likely to break. Lifting from the opposite end, or from a side, makes it more bent- and more likely to break as you pull it out.

Regarding #9: When an insertion has gone badly, The Sensor wire didn’t get “caught on skin” going in. Rather, it’s usually getting caught on the inserter tube while the inserter tube is going out. The key to avoiding this is not to go faster: it’s to go smoother, and to keep the inserter tube perfectly lined-up with the wire on the way out. I find that it’s best to actually pause at the end of the insertion click, so that the device can switch from “going in” to going out" smoothly. (You can feel a funny little bump or two while the plunger is at the bottom, znd you’re changing directions.) Then pull out smoothly, but the most important thing is: keep the clear, outer assembly totally motionless all the way through. When you rush to “do it fast”, you usually bend the inserter assembly a little bit when those “turn-around” bumps happen- and that gets the inserter out-of-line versus the naked wire, actually CAUSING the problems.

Great notes, thanks! Richard, please relay back to Haley, she is free to use anything I said- and she can just incorporate anything she wants to use, with no attribution needed. (If there’s anything of use, she can rewrite as she pleases) :))

Without regard to Haley’s thoughts, here are some things which I know about the topic (and a study reference, if you want to learn WAY TO MUCH about “lag time”, it’s causes, and it’s relationship to device accuracy and noise rejection):

There are biologically sound reasons for ISF-based Sensor readings to have different delays in “rising” bG trends versus “dropping” bG trends, but they result in surprisingly small effects. In rising bG, of course, food gets to your bloodstream first, and must then transfer into ISF through capillary walls. (Except in the brain, Plasma and ISF actually co-mingle in this transfer of glucose and other substances.) But the delay is very short. In falling bG, there is a slight tendency for ISF to fall first, because many body tissue draws on ISF glucose as a source of energy. And (Relatively few cells have direct contact with capillary walls.)

But there’s all kinds of issues going to disrupt this “ISF glucose is the source of cellular energy” concept. First, and maybe most important, skeletal muscle doesn’t use glucose as a primary source! The primary source of energy for muscle power is embedded fat. (Yeah, we humans are “well-marbled” Choice and Prime cuts, not the leaner and less tasty Standard, Utility, or (yikes!) Cutter and Canner grades. Don’t tell your neighborhood cannibals.) Glucose isn’t burned to provide the muscle power; it’s use is more like an enzyme, supporting reaction pathways which generate the bulk of energy from fat. Marathon runners who say things like “I hit the wall at mile 21 because I run out of glucose, and have to start burning fat” have it wrong: The body has nowhere near that amount of glucose, you’ve been burning fat through the entire race.

And second, the biggest consumer of glucose (neurons in the brain) are not surrounded by ISF at all; they’re surrounded by cerebrospinal fluid (“CSF”), and the brain cells which actually touch capillaries grow a unique barrier on these surfaces: The “Brain-Blood Barrier” actually stops plasma from mingling with CSF, and glucose needs to be transported across the barrier via Transporter molecules.The properties of CSF are quite different from those of ISF… although, once the glucose “suckage” is worked back into the common blood supply, the glucose level is shared.

And third, the relationship between plasma glucose and ISF glucose becomes inconsistent during falling bG; especially, during falling bG into the “Hypo” range. SubC ISF becomes relatively more hypo, and it’s glucose is “stolen back” and handed over to ISF within the core of your body. (That’s clever, and kind of weird- a lot like the body sacrificing cold extremities in extreme conditions, in order to keep the body core warm.)

Fourth, more important than the third one-- and maybe more important than the second one, too, because the brain’s energy consumption is pretty consistent: The liver is a huge organ, with a huge blood supply. When I drink alcohol, I’ve found that any uncovered Hypos which occur will hit my brain and bG before they show up on Dexcom: Glucose is being consumed by my liver to process the poison, and those two direct consumers of blood bG are noticably faster than the blood->plasma->ISF->Sensor membrane->Sensor voltage change->Monitor reading path in this case. Fortunately, I don’t drink much, and I always feel a “falling bG headache” in this situation. (And that’s a bit strange, because I’m frequently hypo-unaware in non-alcoholic cases of Hypoglycemia.)

So far, I’ve given you stories. Now for the proof- there’s a study with actual data behind it, finding that the main cause of Minimed “lag time” is it’s own calculations. The time for glucose to cross the Sensor membrane and change the voltage is much shorter, and the delay from blood to ISF is utterly insignificant (in comparison to the effects of their data management mathematics.)

These calculations are done to “smooth out” the Minimed CMGS, keeping it from going “crazy” due to Sensor noise. But they cause long delays, in both directions, and they cause some almost-certain systematic errors as well. (e.g., Minimed will almost NEVER show the true depth of a Hypo episode.) The study clearly indicates ways in which the programming could be improved, although I don’t remember seeing an FDA PMA about an accuracy upgrade. (I might have missed it, I don’t follow MM that closely). This was from late 2007. The persons doing the study had access to all the relevant data: CGMS output, ISIG raw data, ISF samples, bG samples, and Minimed’s programming. Because it was funded (partly) with NIH funds, the full text is available to us. Here’s the study link, on PubMed Central:

Dexcom uses a different mathematical algorithm, and it’s always been “quicker” (especially on falling bG). (That goes all the way back to the original STS, in Spring of 2006.) But it can’t be defined as a single number of minutes. The rate of rise (or fall) is one important factor; the raw data value is another (Dexcom simply has less to work with at low readings); and perhaps most important, the change in slope of the graph-- it’s SECOND derivative, rather than the rate itself. If you are changing directions, it will be slower.

For me, Dex is much later on rising bG (10-20 mins) than it is on falling bG (sometimes right on, usually less than 10 minutes late.) But YMMV, and if I’ve made a big bolus mistake, provoking a rapid rate of change, it will take longer to “catch up.”

Here’s an idea to help you relax about it: If you pay attention and work really hard, your variability will become so small that Dexxy shows a flat line all the time. “Lag time” will be irrelevant!!! (All these years, and I’m nowhere near achieving that, sniff, sniff. But it’s a nice thought, anyway.)

These are gold nuggets of great info. Dexcom does a great job given that monitoring glucose via interstitial fluid is inherently challenging but you definitely have to learn things not discussed in the manual. Hayley does not mention what is obvious to anyone who already uses CGM which is that keeping the sensor taped down is crucial because if the sensor bends within the skin, it will stop working right. It bends as easily as fishing wire and once it kinks, it’s toast. Also, if you are new to Dexcom, keep in mind the readings can lag by 20-30 pts whenever your glucose is rapidly rising or dropping. I found that to be no problem because the lag was consistent enough that I could reliably set my low alert 25 points higher and my high alert 25 pts lower to compensate (this was with previous generation of Dexcom).


Interesting stuff, thanks for the info!

However, I’m very much interested in where your information about muscle metabolism comes from. I guess that’s a polite, and with all due respect, way of saying that it is not the complete picture of everything I’ve studied and learned about metabolic pathways in general and exercise physiology specifically.

An undertaking like running a marathon, as the specific example you used, is going to be driven by muscle and liver glycogen stores mostly, with any glucose taken in during the race acting as a glycogen sparing buffer. Fat plays a role depending on how much output the runner is providing at any given time. Anybody, not just a marathon runner, “hitting the wall” has blown through their glycogen stores. At that point, muscle performance can be maintained up to 25% longer by continuing to take in glucose, then that’s all she wrote.

Most other activities that do not use up to 85% of muscle capacity will, indeed, be burning fat as well. Low level aerobic activities can have muscles burning lot’s of fat, but I’m not aware of too many circumstances where fat is burned exclusively by mucles. It’s actually a fairly easy thing to measure with carbon dioxide and oxygen gas analyzers.

The bottom line? When I’m running, at practically any pace, my Dexcom just can’t keep up with my changing BG. I can drop anywhere between 70 to 125 points in 20 minutes and it won’t even register on my Dexcom. Sometimes, my Dexcom completely misses the drop. It will show a down arrow whiled droppin, but stabilize at a level way above my after run BG low, more closer to a level where my BG is after I correct.

Well- the first thing I see is, the link that I put via the posting tool isn’t in my post! I’ll just enter the study as text:

You are absolutely correct to go after my mention of “a marathon runner”, and your understanding is correct. I was thinking in terms much lower exercise levels, and as you say, glycogen is heavily used even then. That claim of “exclusive” use of fat as an energy source is totally wrong, thank you. (It was too late into the wee hours of AM for thinking properly.)

YMMV- some people experience rising bG after exercise, as long as the session hasn’t been really extreme. (Personally, I’m like you- always dropping if I didn’t pre-load, NEVER experiencing increasing bG as a natural, in-body response.) But your rate of change is very extreme. Do you see “???” and missed graph spots, or does it show relatively level (and wrong) readings?

Np Rick, I got the point about muscle activity in general. It is a really good point to keep in mind and one I often forget because I’m always thinking about exercise conditions.

Thanks again for the info and link!

When I’m running on a treadmill, my Dexcom usually shows readings through my drops. I’ve never seen ??? or missed points just because of running. If my BG is steady when I start running, my Dexcom readings will show steady BG for about 15 minutes into the run, then I’ll get the 45 degree down arrow. Since getting the Dexcom, I’ve never stopped in the middle of a run to check my BG with a fingerstick, so I don’t know when or how far off the readings are when they transition from steady to dropping. By the end of the run at 20 to 25 minutes, I’ve had readings be up to 40 points off, but usually it’s between 20 and 30 points higher than my fingerstick.

And yeah, I’ve had the post workout BG spike many times. I don’t get the extreme spikes into the 200s that I see a lot of people in the forum talking about, but they are noticeable.