Does Narrowing the High/Low Alarm Range AND correcting Within a Narrow Range increase the accuracy of the cgms?

Replies to This Discussion

Permalink Reply by Mommy of Three on June 22, 2009 at 7:48pm

I've been pretty lucky with my Dexcom Seven Plus. It's been giving me readings pretty close to my glucometer. I notice that the first two days I put a sensor in I test 18 times per day (no joke, I have a tendency to want to see if good ol' Dex is on track). I had two different sensors that gave me ??? on the screen and the blood sugar levels ran 30 points or more off - I called Dexcom and they replaced both those sensors for free. A Dexcom rep told me that if a sensor is pretty off like what you mention with the Minimed, to call Dexcom and they will replace the sensor. I always double check when Dex gives a low low reading or a high reading and the weird part is Dex was within 5 points or sometimes exactly on track with my glucometer (I have hypoglycemia unawareness and when Dex starts saying I'm 55 or 50 I sometimes don't believe it and then the glucometer shows me that YEP! Dex was right! I then get my but in gear and chug some sugar - I have a bad habit of ignoring my first low alert which is set at 80 as I sometimes think my blood sugar couldn't possibly go lower but then I forget about the fact that I was just out in the sun in which my blood sugar level will just drop).

Again, I've been pretty lucky with my Dexcom (knock on wood) and Dex has been within 15 points of my glucometer. If it's off 20 points I enter in the blood glucose which calibrates and brings Dex back in line... but again, Dex has been my life saver - especially now that I don't have symptoms when I'm low and my husband can't even tell when I'm low like he used to see in me before. (one of the many side effects of having an A1c between 5 - 6% for 6 years.... ARGHHHHH!!!!)

Sorry my reply is so long.... It sounds like from your post the Minimed CGMS isn't as helpful as my Dex has been for me.

Good Luck!

▶ Reply to This

Permalink Reply by Jan on June 24, 2009 at 9:43am

Congrats on the long term reduction in A1c! Endo gave us a script for testing my niece 16 times a day and we sometimes go over; lately, we have been using all 16. So you basically make sure you "train" each sensor correctly and keep it on track? Interesting. MM sensors were not entirely accurate, but we WERE using them for 9 days instead of the prescribed 3. I could have still used MM sensors to great advantage by watching the trend arrows, etc. MM sensors were painful sometimes even after the EMLA wore off. We were using the sensors only in the arm, there are a lot of nerves there. We use tush area and above for pump. She won't wear pump anywhere else; so that was part of the problem.

▶ Reply to This

Permalink Reply by Don on June 22, 2009 at 8:45pm

I would definitely try the PLUS. Others will chime in, but no I don't think narrowing the alarm range/correcting will increase accuracy. If bg is falling or rising rapidly it is normal for CGM to report numbers up to 30 pts higher/lower than fingerstick. What I like about Dex is that it tends to be consistently higher/lower in this lag, so if falling rapidly it will consistently report #s higher than fingerstick and if bg is rising fast, it will consistently report #'s lower than fingerstick (all within 30 pts). The lag is consistent and allows you to make useful decisions. Set an alert at 80 and you will avert most lows because you can generally depend on the real fingerstick to be even lower so you know to go ahead and ingest carbs. The PLUS statistics suggest it is more accurate than Nav or MM at high range.

▶ Reply to This

Permalink Reply by Jan on June 23, 2009 at 9:48am

Thanks for your response! I gather that Dex 7 Plus is way more accurate than MM; she has an MM pump, that is why we are switching. Plus, MM STILL has that huge introducer needle. So the way to get around that 30 point discrepancy regarding the lows would be to set the low alarm a little higher, 90 perhaps? Very glad to hear Dex is accurate with the highs -- that's more than half the battle!

▶ Reply to This

Permalink Reply by Jan on June 22, 2009 at 10:24pm

I have researched the Dex and I know it is more accurate. We will start Dex 7 Plus as soon as the trainer can train us. I do have two specific questions, if anyone knows: (1) Is it useful to set the parameters on the high and low alarm close in range, i.e. 80 to 130, to increase the accuracy of the cgms? and (2) If you consistently correct when you are SLIGHTLY out of range (for instance if you are 110 and your target is 100), will making these MICRO corrections help improve the accuracy of cgms? I do not like what I have just read about the Dexcom reading lows as 30 degrees higher than they are (that is dangerous, and I have experienced that with the Minimed sensor.) Likewise, I would like to avert a high, long before she gets there. There is someone on the CWD cgms board who has figured out how to increase the accuracy of the Minimed sensors by keeping the sensor's high and low alarms within a very close parameter, (his low alarm is 80 and his high alarm is 110); he corrects BS to 90, even if BS is 110; he corrects (with food) BS of 80 up to 100. He does not correct for 90 minutes after a meal, but at 90 minutes if above a certain range, he will correct. His results have been nothing short of amazing. So I am wondering if there are any adult PWDs who have noticed this and do similar things.P.S. Because of the way this person manages BS with the cgms, BS rarely goes above 180 and the sensor is incredibly accurate.... which is saying a lot, since he uses Minimed's Guardian. Also saying a lot is the fact that he does this for his 11 year old daughter, in puberty, which is a very challenging time period for keeping blood sugars stable.

▶ Reply to This

Permalink Reply by Rickst29 on June 23, 2009 at 10:39am

Person on CWD is making a bad guess about WHY he gets high accuracy. Alarm settings are of no significance at all; he achieves good accuracy by (1) maintaining a very tight bG; and (2) doing all of his calibrations within that tight range.

When your calibrations are all done within the range of 60-150 mg/dL, AND your entire life is spent there too, you'll of course have much better accuracy than someone who enters calibration data at 180 mg/dL, and another entry at 140 mg/dL .... and then expects that the line which Minimed draws between those two values to establish a calibration "curve" will be accurate at far lower bG levels. (And MM does use a straight line calibration formula, not any sort of fancy "curve".)

It sounds like that person on CWD is doing well, although I would not wait 90 minutes before performing a meal correction. I know how nearly all of my meals and insulin combinations affect bG over time, from the insulin dose (right through the meal and right through the next 4 hours. For me, any significant deviation from expectations calls for treatment immediately, either extra carbs or extra units.

That's the advantage of having a CGMS with a graph... If he's not going to look at the trend in the graph for 90 minutes, he's really just treating it as a pain-free fingerstick monitor. It's far more useful than that if you use it correctly.

Treat the TREND, not just the bG. (BTW, I have the same name over there as I use here.)

▶ Reply to This

Permalink Reply by Jan on June 23, 2009 at 7:10pm

".....Alarm settings are of no significance at all; he achieves good accuracy by (1) maintaining a very tight bG; and (2) doing all of his calibrations within that tight range..."

Yes, Yes, you have got it. Maybe I phrased it incorrectly. So you agree that within the above parameters, which are "maintaining a very tight BG and doing all of the calibrations within that tight range", that it is possible to use the cgms in this fashion?

Re the postprandial spikes: If you see a high at one hour you would treat with insulin? How high do you allow yourself to go before you treat? And, obviously, you are going to have to feed that IOB in a few hours. I will certainly make use of that tip (we do overbolus at times but only if we are able to watch her very carefully). She could not use it as I believe she could ignore an alarm or if, for some reason, the alarm did not go off and she forgot to eat extra there would be a problem.

▶ Reply to This

Permalink Reply by Rickst29 on June 24, 2009 at 2:05pm

Exactly! If you're living between 80 and 140 "nearly all the time", you want to do your calibrations at or near that operating range. If you add calibration data DURING incidents of much higher BG levels (which you experience only infrequently), you are likely to increase errors within your "working" range. In the low range, Dexcom raw signal levels and accuracy are so compromised below BG of 60 mg/dL that you shouldn't calibrate there.

Give it good data between these endpoints. More than 140 is OK, if you frequently are over 140. But although Dexcom IS willing to take calibration readings down to a very low level of bG, I think that readings taken below about 75 don't help-- I suspect that it's at least as good, maybe better, to simply let it use the curve which has been established using data points between 75 and 120.... the math is probably less error prone than the recalibration data.

But that's my SWAG, and Dexcom clearly disagrees with me on this. They're willing to accept meter calibration readings for much lower levels of meter-measured BG, and it's certainly possible that their choice is smarter than my SWAG.
- - - - -
Postpriandial spikes: I add more insulin immediately after I see (and confirm) high readings which I didn't expect to get. Even during the meal, and often as little as 30 minutes post-prandial. However, "expectations" are really complicated:
1) how many bolus units, of course.
2) what period did I extend the bolus over?
3) how much "lead time" did I use in starting that bolus before hand (if any)?
4) and why did I use that particular lead time amount?
5) is alcohol involved, and if so, did it add lots of carbs too? (e.g., beer is high carbs, slighyl less alcohol, but bigger servings; dry red wine contains a lot more alcohol versus the carbs it contains, with a much smaller serving size too.)
6) Order of the meal (If wine came first, WITHOUT plain carbs to "assist", or salad with cheese and oil came before carbs, the carbs come on much slower. lower BG in the first 90 minutes is EXPECTED, but it will pile on later-- sort of a smaller version of the "pizza effect".)

Meals I control are slow and even to digest, I usually don't exceed even 140 mg/dL. If the curve is showing sharply upwards AND I can figure out that I made an insufficient bolus, I'll stack on some more even as low as 120 mg/dL. It's all about the plans and expectations versus actual results.

But, in unfamiliar restaurants, you don't know exactly what's coming, or when it will arrive, until it shows up. I'm often above 140 for a while because I go more conservative. My home-style treatment would often lead to falling bG, due to a slightly slow kitchen, and I'm not happy with ruining my appetite from "meal" of sugar tabs before the plate of $$$ food even shows up.

Being protected by CGMS allows a radical change in lifestyle: You can be ultra-aggressive with insulin, because it's easy to back-fill when you see your bg curve turn downwards. With large amounts of bolus insulin on board, I'm often treating "lows" at bg of 110, even 120. Because I can see, with almost absolute certainty, that's it's headed lower than I like. That oncoming low doesn't even happen-- I didn't just treat it, I AVOIDED it. :)

For me, it's "Shoot up early; Shoot up big; Watch Dexcom like a hawk while bolus IOB is still active; and back-fill early, as soon as I see an unexpected or too-steep drop-off in the shape of the curve."

That's for me. But for her, you've already made the right choice: If she "could ignore an alarm", she should not try to do as I do, it's dangerous.
Heck, as you just read, I try not to ever let "low" alarms even occur: I'm already "back-filling" with glucose tabs at 110+, it's absolutely not supposed to even reach the 80 mg/dL alarm level when digestion and bolus are active. If she's likely to ignore alarms, she's certainly NOT going to "watch like a hawk" when there aren't alarms.

And A1C-wise, my "aggression" is about the difference between 6-something and mid-fives. It's not needed to move into a performance range of "high-quality treatment results" which would satisfy any MD -- number well below 7.0, or (in the case of my own MD) below 6.5%. You get there by avoiding the really MAJOR screw-ups (Doses totally forgotten until your bG "blew up" in your face; nibbling un-measured quantities of carb-heavy foods; going to bed with food, or insulin, or BOTH still on board; stuff like that.)

▶ Reply to This

Permalink Reply by Jan on June 24, 2009 at 10:32pm

I am definitely going to print this out. We do do a version of the same with her dinner bolus, even without cgms, catching IOB at the 2 hour mark. Now she frequently will spike 200 and over, but I think with cgms we will be able to catch more of these. It is imperative for us to catch BG rise before it has time to occur; a rise of 100 points can happen in as little as an hour. I'm sure it is growth hormone related. Or we just switch her basal; it can vary, night to night slightly, but enough to make a difference. Her pattern usually remains the same though. When she is not under our care, we have to settle for having her BS at 200 2 hours postprandial; at the end of four and a half hours she will be down to 80. At home, because we manage everything, we can certainly try the above. My main hope is that Dex is accurate for her and that it is not too much of a pain, literally and figuratively, and that she will agree to wear it for a large percentage of the time.

▶ Reply to This

Permalink Reply by Rickst29 on June 23, 2009 at 10:24am

The alarm settings have nothing to do with the accuracy of readings. They only determine when the Dexcom starts buzzing and shaking, and generally NAGGING AT YOU to do something. (My own Dexcom settings are 80 and 140, but you seem to be speaking of another person-- if it's your child, or any child, those settings are too risky-- too late at catching lows, and probably disrupting life with "high" warnings after every single meal.)

With CGMS, you correct according to the TREND, even more than the reading. "Correcting" a 120 when you don't know the trend direction is definitely a mistake-- you need to confirm that the trend is UPWARDS, or ABSOLUTELY LEVEL, before doing anything with a 120 mg/dL reading. Since insulin is so slow, 3-4 hours for it to finish working, treating highs before you're absolutely sure that it's the right thing to do is likely to cause those roller coaster swings, rather than prevent them.

Example: Suppose I treated at 120 before confirming that the trend is upwards, or absolutely stable. Then, over the next 20-30 minutes, (which is BEFORE the insulin really even gets started, I see that I'm trending downwards already-- and I'm already around 100, or 90, and the insulin's activity hasn't even started yet. It's time to back-fill, that dose was definitely a mistake.
- - - - -

Fine tuning treatment of lows is WAY easier, because glucose (tabs or gels) works in only 10-30 minutes. And less expensive "regular" carbs, such as white bread, are also completely done within 40 minutes (if they haven't been loaded up with fat or protein content). If I'm below 80 I'll grab a tab or two-- maybe 3, if the trend shows that I'm falling sharply. Then, if Dex shows a sharp turn upwards upwards after about 15 minutes, I stop treating-- let that first treatment FINISH it's job, and see where it ends up, and whether I'm now tending a flat line. If I'm still falling, or if the bG "jump" merely went level, then I need more-- and I watch that graph like a hawk until it's good.

When driving, of course, I load up at even 90 mg/dL. I pull over and park the car if my 80 alarm goes off. Driving the car, I aim for 120-150 mg/dL.
- - - - -


Are you absolutely sure that that you want to TREAT at 120 mg/dL reading in level trend conditions? That's radical-- we're talking A1c readings in the 5s, maybe even 4s. But there's one thing here that you seem have missed:

With a reliable CGMS (obviously not MM in her case, she needs to try Abbott or Dex), you don't wait for the alarm and ignore the trend. If the Dex is recently calibrated and confirmed trustworthy, you treat the trend FIRST, and the alarm probably never happens. example: I'm doing yard work, and I see a downwards trend developing (114, 108, 102, 96 over the last four readings. Do I actually wait for it to reach 80 and buzz before I eat something, and have to stop working while I wait for the sugar tabs to "catch up" with my yard work?

Of course not. If you've got a clear trend towards trouble, treat it immediately. Trouble can be prevented, if you use the trend line properly.
- - - - -

[Gentle-Slap]
I'm bothered by your blanket statement "I have researched the Dex and I know it is more accurate." (Even though this WAS stunningly true in my own case.) For many OTHER people, MM accuracy is excellent. Switching brands often allows the pt. to get more reliable readings, but I know of instances in which the successful switch was BACKWARDS from your blanket claim. (Dex sucked, and MM didn't.)

In brand-swapping, Abbott seems to have the best rate of satisfaction on accuracy-- but falls on it's face with customer service and hardware reliability (at least so far). Dexcom seems to have the best customer support, and a rate of accuracy success which seems almost as good as Abbott's. MM definitely seems to have the highest rate of "bad accuracy" outcomes, but it works really well for many people.

Dexcom's Transmitter is way better than the others, especially the Minilink. (Minilink is costly, tends to die young, has a horrible hassle with frequent recharging, AND sometimes just goes "bad" for no obvious reason at all.) But Dexcom's "can't replace the battery" monitor costs a big chunk of change to replace, and it also leaks like a sieve. (Guardian and R/T have pretty good resistance against accidental drowning, while Dexcom has absolutely none.)

Each product is it's problems and issues. And in the accuracy area, saying that Dexcom is more accurate than Minimed-- for EVERYONE, and just from reading a lot of study data and posts like this one, is unfair and not true.
[/Gentle-Slap]

▶ Reply to This

Permalink Reply by Jan on June 23, 2009 at 10:54am

Rick, I have spoken to you at length on CWD chat. Hoping you would chime in as you are a true expert. We have not yet started Dex (trainer keeps changing appointments; hope she will do it next week now). The question does sound odd. On the CWD boards there is a parent who has managed to get around the glitches in the MM cgms by keeping his DDs high/low alarms set to a very narrow range and correcting highs/lows within that very narrow range. Seems by keeping within this narrow range MM's cgms is extremely accurate (parent is a computer engineer). He finds using his technique he does not even have to do a fingerstick to confirm before dosing. I think his DD does two or three fingersticks a day. Could you go over to the CWD cgms forums, read the post and let me know what you think? He sets his low alarm to 80; high alarm to 100, I think. If DD's BS is 80, she immediately eats 4 grams or so of carbs to bring BS up to target of 100. If her BS goes over 100, she immediately corrects (without fingerstick) to 90. The kid is in a nondiabetic range 70 percent of the day with A1cs in the 5.4.range, since he started implementing his technique four months after dx. He says no honeymoon, but I think her TDD indicates otherwise; I'm sure his system, used four months after dx has preserved a lot of this kids beta cells. I would not dismiss his findings out of hand. I think he may be on to something. I am up all night with my niece's BS 5 nights a week, Mom two or three, so she is covered 100 percent all night long. And it is a challenge to keep her in range at night. We usually have problems at some time during the night. Basal testing means nothing; as every night is essentially a basal test and she varies (albiet usually within a semipredictable range) nightly. I plan to try his technique, but I guess I'll confirm with fingerstick. Believe me, she is getting a fingerstick almost every hour a lot of nights anyway. His daughter is 11; my niece is 12. I think this guy's findings may be able to be applied in various ways. P.S. His DD expects an alarm every hour and responds to it and does not mind. My niece most certainly would mind so I can only try the man's technique when I am managing her D (nights and if she is hanging out at home on computer, homework or TV). Fortunately, at the time of her greatest instability 7pm until 2 or 3am, she is home doing such. She's pretty stable during the day. I know you are certainly qualified to answer my questions, particularlary if you read the post. P.S. Using my technique of controlling BS, fingersticks and aggressive corrections plus carefully watching, we have been able to keep a child long out of honeymoon, with fluctuating basals, who is in puberty in the 6.1 range most of the time, although we have had 6.4, 6.7 and even 7.0 within the past year and a half. I KNOW it is possible to get a child in the high 5s with cgms, at least I know I should be able to do it with cgms. PPS Endo's goal has been, and continues to be for us to keep BS in the 80 to 110 range nightly. We really have not been able to do this reliably without cgms; either she will spike and we will be correcting and watching a high or she continues to drop or slightly rise on her basals (basals are tested nightly). She usually will vary within two or three nightly patterns, i.e. .95, 1.0 or 1.05, with occasional forays in the dropping like a rock; spiking high categories.

▶ Reply to This

Permalink Reply by Rickst29 on June 24, 2009 at 2:44pm

Sorry, I must go totally off topic for a moment, before responding to your "real post". I love getting your "kudos", but I must take a moment here to emphasize for EVERYONE that I am totally unqualified to give medical advice, or "support" advice on any of the CGMS models. My opinions, experience and feelings must be treated for what they are-- unqualified, nonprofessional opinions. I have never possessed a medical license of any kind.

The problem word was "expert". Why do I have to go wild about that word? It's because my DW is a licensed MD in an unrelated specialty, and I am an officer within her medical practice Corporation. "Loose" wording which includes legal terms, no matter how well-intentioned, can create career-threatening problems for us. (The problem here is that "expert" is well-defined legal term within the Medical world, e.g., when "expert witnesses" give testimony in malpractice lawsuits.)

Call me instead a "user of great experience", or "CGMS user of great experience", or something. Thanks! I'll respond with my D-related and CGSM-related opinions and feelings ;) about what you said in another post, right now. Thanks.

▶ Reply to This

• First
• Previous
• Next
• Last