A person fielding my call at Medtronic stated today (7 September 2012):
"The most accurate and reliable way to check your glucose is with the interstitial sample such as when using a CGM, not with a home blood glucose meter with a test strip."
I find this difficult to believe unless there is some other explanation. It seems as if the conventional wisdom is just the reverse of what the pump/CGM rep said. I thought CGM's had a lag time and also a 10% bigger margin of error.
Then I read two old comments where folks reported an endo said the same as the rep I quoted. Can this be so?
Last time I checked my impression was people with a CGM, with or without a pump, do NOT rely much on CGM numbers but still use a traditional home meter often to check bG since they feel that interstitial G is not as reliable as blood G.
I want to learn the explanation.
Also, I am under the impression that Medicare in US does NOT pay for CGM units (nor it's costly supplies) since the data is not as reliable as the old fashioned typical blood glucose meter.
It's wierd. I sort of try to make my CGM and BG meter line up as it makes the CGM work better. At the same time, if I have any doubt in the CGM, I'll double check it with a meter. I mostly always test for meals with a meter too. I would have asked him to email me some proof of that as I'd be intrigued to see if he could come up with anything.
That being said, I've improved my A1C a bit since I started using the CGM, although it's a small sample size.
This may be a stretch, but I wonder if it's relative to what happens in people w/o diabetes. Does the pancreas, liver, hormones etc that 'recognize' glucose levels, rely on glucose in blood, or tissues, or both ?
Just because we use blood glucose measurements from meters (which is estimated plasma reading, from whole blood), to determine insulin dosage, doesn't mean that's the 'right' number. (but it's a huge improvement from testing glucose in urine in the dark ages !)
So maybe the rep was saying that glucose level in interstitial tissue is a more accurate representation of how the human body senses and responds to glucose level. But for us humans using CGMS, our point of reference is blood glucose, so their algorithms and calibrations have to 'translate' it to a number that we have been trained to use.
If we/CGMS were truly able to 'think like a pancreas', would we be more interested in knowing blood glucose or ISF glucose ?
to your second point.. I have used both MM and Dexcom CGMS, and you are correct. We are trained to dose insulin only on meter BG, not CGMS. I think the CGMS does accurately 'sense' ISF glucose, but the 'translation' to blood glucose is the challenge. This is why there is emphasis on the trends being more useful than the actual numbers.
And you are correct that Medicare does not cover CGMS, although some have supplemental plans that cover it.
If that were so, why is it then that (a) they are calibrated with a fingerstick (blood glucose) and (b) in their approval testing, they are compared against a lab blood glucose meter (and not the other way around).
Finally, if you look in the manuals for a CGMS, such as DEXCOM (the other are no better), you will see tables documenting how inaccurate they are and how often you will get false positive and false negative alarms.
Don't get me wrong, I htink they are incredibly useful tools.... and I wear one 24x7x365.
But don't be fooled that they are more accurate or reliable than even a $20 drugstore BG meter. They are not. And this is objectively shown by the data in their own manuals and in their approval tests.
Until the current CGMS sensor include separate absolute oxygen sensor to accurately check residual oxygen, the meter has to be calibrated with the caveman fingerprick machine to correct the sensor to what fingerprick is reading.
It is really hard to see any validity of accuracy versus caveman machines when in fact one needs to use fingerprick machine to calibrate residual oxygen to actual BG reading at fingertip at regular intervals.
If you are using the MM sensors, there is a way to tell about how accurate your sensors are from one of the Carelink reports - 'Sensor Daily Overlay'. There is a field on that report called 'MAD%' - stands for 'mean absolute difference'.
It is the difference between the sensor reading and the BG you enter to calibrate with. If you can achieve a MAD% of around 10%, you can consider your sensor readings to be quite accurate. I've often bolused based solely on my sensor readings (although everyone says not to do that!).
I think a normal MAD% hovers around 18% - which would make it a little dicier.
This is not rocket science. A CGM reads interstitial fluid. A BG is capillary blood. There is a statistical correlation between the two. One is no better than the other. They work together as a tool system when they are calibrated together by measuring capillary BG and entering it in the CGM. Pearl: Dose all insulin from a finger stick BG.
Medicare uses HCPX codes and has not approved the ones for CGMs yet. AARP is working on this I have been told.
For me, the dexcom CGM was very inaccurate. I actually found myself using MORE strips while using the CGM, just because I was constantly looking at it and thinking "that reading makes no sense" and then checking with strips just to make sure that the meter was wrong. The margin of error for me far, far exceeded anything reasonable. I've seen it say I was below 40 when I was actually over 160. Seemed to very often alert low (below 40) while I'm sleeping.. but I take no long acting insulin, so I've never been low while sleeping, and it would be essentially impossible for me to be.
Even with all the faults I've just listed, I can say, as others have, that the trend is usually accurate.. and that when I was using it it did lead me to correct for several moderate highs that I would have otherwise been unaware of, and was what prompted me to actually check with my real meter to find the lowest BG level I've ever had (39).
The blood exchange between the interstitial layers and the blood circuit is slow. The levels in the interstitial areas are lagging 15 minutes behind. Depending on the relative movement of the glucose the number on the CGMS can be very misinformative although the visible trend will help to understand the outcome. So you want to eat and the CGMS shows 90 with a trend down. You measure with the meter and the number is 66. Obviously there is a big difference in treating these two situations taking the numbers alone. It is like changing the glucose level in the amazonas and mesuring the glucose in its mangroves. It takes time until the differential in glucose has lead to diffusion processes to equalize the differential. So CGMS in the upper layers of the skin will never be as accurate according to timing as glucose strips. It is still a great tool to not oversleep lows. It will also help to correct highs earlier by waking you up at night.
In comparison to CGMS and meter the pancreas has very direct access to the blood stream coming from the digestive tract. The beta cells directly respond to the level of glucose by releasing micro amounts of insulin. Thus they can quickly react to the increase or descrease in glucose levels. At max the healthy amount of beta cells are capable to produce 20 IE per hour.
In addition to the issues that Holger discusses regarding the diffusion delays between blood and interstitial glucose levels. there are errors due to the physics of the sensors themselves.
Today's sensors are mostly based on measuring electrical currents created by the sensor (amperometric). These have at least 3 sources of error.
1) Calibration error. These currents are created by the reaction of glucose with oxygen or with a chemical on the sensor . When you calibrate the sensor, you are letting the device know what current level corresponds to what blood glucose level. The CGM then uses how the current changes to signal how the BG is changing. Not all the electrical current is generated by the reaction with glucose - some of it is due to other factors, including the variations in oxygen mentioned above. Also the sensor currents may not be linear with blood glucose...it may change more or less depending on the BG level.i.e. the sensitivity may vary.Sensor delay
2) Sensor delay. In addition to the delay caused by the diffusion of glucose from the blood to the interstitial fluid, the sensor itself may introduce additional delay. This is caused by the averaging and other signal processing that the CGM does on the raw sensor signal... i.e. it may take several samples of the sensor current for the CGM to priduce a BG number. Also it takes time for the interstitial fluid to diffuse across a membrane in the sensor and react with the enyzymes in the sensor to generate the current.
3_) Sensor drift. The currents produced by the sensor can drift over time due to numerous factors that are not well understood. Some of the reasons include changes in blood flow, oxygen levels, blood vessel constriction and expansion, and the inflammation that the sensor may cause in the body where it is inserted. This can cuase the body to mount a reaction where the immune systems mounts aresponse and sends white cells to the site of the sensor, which in turn consume oxygen and distort the sensor readings.
I am a mom to a wonderful 8 year old boy who has diabetes. I’m always impressed with the level of knowledge and care shared on the site. Thank you- I’ve still got a lot to learn (we’re a year +3 into the process). We actually have a CGM and I can’t convince him to try it. It would help us to understand his nights better with less worry maybe later on but we’ll have to wait… Pumping since February and another thing attached to him is too much for now.
This is a most interesting and revelent post about cgms.
I am T2.
T2 diabetics make up 80 to 85 per cent of all diabetics.
One would think if one had their marketing hat on; there woud be a desire to chase the T2 market place based on numbers and provide a machine that also helps drive the T2 issues. Dexcom basically told me I could drop dead.
Dexcom does'nt and loses money every report. Get the price down, get the volume up, make it easy to use and make sensors more reliable -longer running period.
Nope, we are driving T1 and in my mind the machine and its alarms are targeted around a properly/new/normal person and the T1 goals of driving it like it was on normal person not a T2 running sub 7.0 a1c.
If that was done, the unit could become very popular and accepted by insurance/industry and in fact its trends and pictures are truly helpful policing the eating cycle/diet and body operation.
In terms of the electronic industry history it is akin to the old Ampex large studio video machines sold in tiny quantities per year when Sony and JVC were out to make zillions of machines at a couple of 100 dollars and falling.
History is complete, Ampex died, Sony and JVC made millions and we all got the benefit of cost effective home video tape recorder in its day.
To be fair; one has to concede we have the FDA and government ham fisted control and ideas in this loop and that may be throwing out the style of agressive cost effective development and improvements shown in the PC Industry that I believe could truly benefit the diabetes problem.
Maybe if we are lucky; somebody in Asia may take up this challenge. The numbers of T2 is huge world wide.
Dexcom and Type 2's is an issue of Insurance Companies not reimbursing for Type 2's. Dexcom gladly sells to Type 2's but they can't give it to you for free. If you listen to Dexcom's last conference call, they speak directly to this issue and how they feel Insurance Companies are slowly becoming more accepting of reimbursing for some Type 2's. You are right if you add all Type 2's into the market it will be huge.
