CGMS ISIG and Calibration Guidelines

This post is to explain how to calibrate a CGM effectively to achieve overall good accuracy.
Also included is a description of the CGM's ISIG indicator and how to use it to guage confidence
factor for each calibration.

Calibration
When you calibrate a CGM, you are telling it the current BG. The CGM compares the BG you enter
with the sensor's electrical output (known as the sensor's ISIG). Let's say that you enter a BG of 120,
and the sensor's ISIG at that moment is 12. The CGM then knows that the current ratio of BG to ISIG
is 120/12 = 10, so from that point forward, all future sensor ISIG values are multiplied by 10 to create
the on-screen SG (sensor glucose) display.

For example, if 10 minutes later the ISIG has increased from 12 to 13, the CGM's SG reading will
increase from 120 to 130.

Bad calibrations happen primarily for these reasons:

1. When there is an inaccurate BG reading (such as when fingers are contaminated with sugar)
2. When BG values are very high or very low. The sensor's ISIG at extreme BG values can not
   be accurately extrapolated to the normal BG range.
3. When BG is changing rapidly. This causes the BG you enter to correspond to an ISIG that
   is delayed in time.
4. Sensors near the end of life. Near the end of life, the sensor's sensitivity declines by the hour,
   so the BG-to-ISIG ratio is not stable. A calibration with an end-of-life sensor is good only for
   a short time period, if at all.

Therefore, the following practices help to insure a good calibration:

• Make sure hands are completely clean and dry before the BG. This is always important, and
  especially important when relying on the data to calibrate your CGM.
• Avoid using BG's under 70, or over 140 for cal's.
• Only use a BG for a CGM cal if the CGM shows that BG has been relatively "flat" for the past half-hour
• Never cal right after you eat. BG is already rising 15 minutes after you eat.
• Never eat right after you cal. The CGM is counting on your BG remaining stable for 15 minutes.
  (In other words, avoid eating both 15 minutes before and after you cal, if possible).

If you have no choice but to cal under poor conditions in order to keep the sensor from timing out,
be sure to do another BG test and cal as soon as BG stabilizes again.

ISIG
The ISIG (short for Insterstitial Signal) is an electrical reading that is proportional to BG. In theory,
the ISIG is linearly propoortional, but in practice it is linearly proportional over a limited BG range,
which is why you always should cal when BG is within a normal range such as 70-140. Cal's at 50
or 300 might not linearly extrapolate into an accurate reading when BG is in the normal range.

The ISIG provides an additional tool to gauge confidence for each calibration. On the Minimed Guardian,
ISIG can be read by pressing the ESC button twice. Most other meters should have a similar option
to view the ISIG.

To make use of the ISIG to improve calibration confidence:

• Each time you cal, look at the ISIG value at the time of the cal, and determine the ratio of BG/ISIG.
  For example, you may find that a typical ratio is 15:1, or 8:1.
• For the lifetime of your sensor, the BG/ISIG ratio will remain relatively consistent, but it will change
  somewhat from cal to cal (which is why you have to do cals). However, if your sensor starts at a ratio
  of 12:1, it usually will remain in that general vicinity during its useful life.
• If you do a cal and find that the BG/ISIG ratio is substantially different from prior cals, it is an indication
  that something might be wrong with the sensor. For example - Let's say you usually have a ratio of 12:1,
  and then one cal has a ratio of 5:1. This is a suspicious cal. Check your sensor to see if it has loosened,
  or if maybe it has been subject to physical pressure such as sleeping on it, or if it has been in use for it's
  typical expected lifetime.
• Another possibility when you see a suspicious BG/ISIG ratio is that BG just started to change rapidly
  around the time that you did the test. Watch the CGM reading over the next 20 minutes. If you do see
  a rapid change, cal again as soon as the BG stabilizes.

Finally, if you get a BG reading that differs dramatically from the CGM, don't jump to conclusion that the CGM
is wrong. It might be a contaminated BG reading. Always re-check the BG and don't re-cal the CGM until you
are certain the the BG is correct, or you may turn a good cal into a bad one.

There is always a possibility that the CGM will be wrong and you'll have a low or high bad enough to be symptomatic.
Not nearly as often though as relying on BG checks alone. The key is to follow good calibration procedure, and
to use your judgement at all times in interpreting the CGM data.

The incidence of false CGM readings can be greatly reduced using the methods above.

Here's an additional resource with even more complete information: http://www.myparadigm.eu/

Edited 8-10-2009 - A tip to get good initial calibrations with the MM CGM - install the sensor at least
2 hours before attaching the transmitter. We do this by installing the new sensor at night, then switching
over the transmitter in the morning. This has helped get more consistetly good initial calibrations.

 

Wow. That's some great information, Darryl. Thank you so much for posting that. I'm printing it off and keeping it in our binder for future reference.

I have a question for you - due to an unfortunate snack choice by Grandma, we couldn't calibrate before supper. Our calibration was due by 7:30, and so we haven't had readings until I checked him about a half an hour ago. His sugar was 182. I wasn't thrilled with that, but I used it as a calibration not only because I want readings during the night, but for the following reason. Let me know if this makes sense:

Our Medtronic rep told us that there is a "sweet spot" with respect to the relationship between ISIG and the BG numbers. The actual algorithm is patented, of course, but she told us that if you divide BG by ISIG, the closer that number is to 8, the "better" the calibration. That doesn't mean that if that number is 6 or 12, it's a bad calibration, but she did tell us not to use a particular BG if that number was above 20, and again, the closer to 8, the better.

So...when I divided his BG of 182 by the ISIG, I got a result of 7.8889879 or something like that. Because it was quite close to 8 (and because I was impatient to get readings again), I used it as a cal, even though it was higher than I liked. Does all of this jive with your understanding of calibration and ISIG? Would you have still waited for sugar to come down more, or would you do as I did? (I know you're in a lot tighter control than we are, so you probably don't have this problem very often, but your opinion would mean a lot to me.)

Thanks a lot for your advice, and thanks again for all the time you took to post this. Very informative!

 

Well, unfortuntately I think that is not exactly correct.

First, we've used 70+ sensors, and rarely have seen an BG/ISIG
outside the range of 9 to 15.

Second, you have no control over the BG/ISIG.

I think what the rep meant to say is that if the ISIG is way above
or below 8, it is telling you that the BG is out of the normal range.
Although again, in our case, a ISIG of 8 would typically correspond
to a BG between 50 and 80.

Another possible valid interpretation of the rep's advice would be
when BG is changing fast. In that case, the BG/ISIG ratio would
be different from normal due to the time delay. That is essentially
similar to what I described in part of my post.

In any case, don't get hung up on expecting a ratio of 8. You will
have ratios different from that and it does not mean that they are
bad (nor does 8 it mean it is a good cal - if your ratio is 8:1 at a BG
of 180, it will still be approximately 8:1 at a BG of 250, which is
clearly not a great BG to cal at). The most important thing is to
know the approximate ratio for each sensor, and to look for
departures in any given cal from the ratio you expect.

Maybe you could run this by your rep and see if they can clarify?

 

This website has some extremely interesting information about BG Values and ISIG with the MM CGM. In some ways, the technology has a ways to go, based on this information. But, the cgms is extremely useful and is a godsend. With a larger sample size of users, the technology will be perfected.

http://www.myparadigm.eu/

Thank you Darryl for the tips!

EDIT:Especially Look at Sensor Instrumental Delay and Behavior of the Newly Inserted Sensor

 

Thanks! I think you sent me that link before and it has a lot of great info and data.
I will edit my original post to include the link.

 

Thanks Darryl for the great tips. Thanks also for putting links to your posts in your signature. I will definitely use them for reference once Jacob has a CGM.

So how about the importance of using the most accurate BG meter possible for calibrations? And how can one be confident in the accuracy of their meter?

 

That's a good point... when we researched meters, the one touch and freestyle
seemed to have similar and very good correlation to BG in the plots.

The Omnipod PDM has a built-in freestyle meter, so we use that.

I'm pretty sure all the current home BG meters are good enough to be relied
upon as long as the test site is clean.

 

It does matter what meter you use while calibrating. The more accurate the reading from the fingerstick, the more reliable the cgms will be. I think the aviva is the most accurate meter. The most accurate meter will most likely give the most consistent readings. Also, if doing multiple blood tests on a certain meter at the same time, the least amount of variance between tests helps determine this factor.

Watch this dexcom user on youtube. He is using the same sensor, with two receivers calibrated to different meters. The receivers use an algorithm based on the Fingerstick BG readings to correlate to actual BG/Interstitial Fluid Glucose reading. MM has the algorithm patented, in fact. I could provide the papers, if I had to. There is a HUGE difference between readings on the receivers. He talks about all of this stuff in about the last minute of the video. http://youtube.com/watch?v=gGgfhcxD5FE

 

That is a very interesting video.. I didn't even know you could use more than one received with the same sensor but I guess there's not reason why not!

I was wondering though, the two BG tests he did came out 176 and 179 (then he forgot the 2nd result and programmed a 175 into the 2nd meter). He showed that the meters were off by 15 points (= 0.5% A1C, which is indeed significant), but yet the two BG's in the video were off by only 3 points.

If the difference is really 15 points, though, it certainly could matter to A1C.

Monica, do you know a source for any current BG test meter accuracy studies?

 

Unfortunately, no. I could probably look it up tomorrow, since I have school and access to journals there.

But remember, the cgms relies a lot on the meter calibrations. I bet the maker of the video (Bernard) is probably right. He knows what he is talking about.

Also remember that the transmitter is just sending a signal. It is not "smart", and if the receiver has the serial number for the transmitter, it will receive the transmitter's signal...even if there are two receivers receiving the same signal.

 

I just found one study about a minute ago from the ADA, comparing four meters. Very Interesting. Has charts.

http://professional.diabetes.org/Content/Posters/2004/p439-P.pdf

I will continue to look for more studies. I want to get a study with the Aviva meter.

 

The article points out the test strip accuracy degrades with time... I did not know that.
We have a lot of half-used bottles around for convenience, so maybe that's not a good idea!

Here is a good study on the base accuracy of the Freestyle vs. the Accucheck. You need to
register with Medscape but it's a simple and free registration, worth it to get the article:

http://www.medscape.com/viewarticle/541546_print

.

 

I get some "off readings" with the flash. What's scary is that the cgms I use is more accurate than the flash. In fact, sometimes when I test, the flash is 50 points above the actual cgms reading. I then retest, and the number correlates within one-five points of the cgms reading and will be 50 points lower than the initial fingerstick reading. I have used the flash prior to cgms for a little bit, and I quit using it due to inaccuracy. But now I use the flash for calibration, etc. It's weird that it can be this inconsistent. It's probably the heat, too even though I avoid the heat as much as possible.

When I get a reading that is off, when I am trying to calibrate, I just mark it as a "control solution" check, and the cgms ignores the inconsistent reading. Plus, the navigator cgms will reject an off reading anyways.

 

I understand the Navigator will not allow a calibration if bg is rising or falling. So it's one less issue.

 

Excellent information, thank you very much. I just wonder if my son's BG is always rising or falling and never stable.

 

We use the MiniLink for my daughter, and have used it 24/7 for a year and a half now. We get excellent results, generally, our meter readings and sensor readings are within 5 to 10 points of each other. Two week averages are almost always within 5 points. Calibrating correctly is so important.

I thought it worth noting that if you can't calibrate, don't. You can always put one in later when BG's are stable. Also, when you tell the system that the BG is 120, you will not necessarily see 120 on the screen. It averages the last four or five readings together (and ISIG's) and determines the value.

One more thing, when you restart (speaking MiniLink here) on day 3, you are starting over with a clean slate. The system does not remember past calibrations. For this reason, if you see a lot of really whacky numbers, you can try to restart the system with a good, stable BG and try again.

 

I learned from one of the MM support techs recently, that it is really important that you only calibrate 3 to 4 time a day and no more. I think when I initially read that in the book I interpreted it as at least 3 to 4 time per day. So, for a long time I'd been calibrating any time we checked BG and Sean's SG readings had been relatively flat over the previous 15 or 20 minutes.

I'm finding that now that we're only calibrating 3 or 4 times a day, we do indeed get much better results with the CGMS.

Does anyone here know enough about the algorithm to explain why calibrating more often (but still only when stable) would yield less reliable CGMS readings?

 

It does not matter if you calibrate 3 times or 30 times a day, what's important is how accurate the cal's are.

It so happens that most people who would calibrate more than 3 times a day are, by necessity, including some BG numbers
that not appropriate for calibration. These would include BG's outside of the target range, or BG reading taken when BG is
changing quickly.

So, it's not that more than 3 cals is a problem, it's that people who cal using every BG reading - rather than being selective -
are likely inserting some bad cal's into the algoritm.

The minimum number of cal's I recommend (based on 18 months of successful experience using the MM Guardian) is 3 cals.
The best time to cal is before breakfast, before dinner (at least 2 hours after the afternoon snack), and at or after bedtime.
These times are spaced at roughly 8 to 10 hour intervals, at times when BG is typically flat.

However, any BG not meeting the criteria in my post CGMS Calibration should not be used to cal the CGM.

However, regardless of when cals are done, 3 good cal's a day are certainly sufficient for reliable and accurate CGM data.
More cals than that in a day does not hurt (as long as the BG readings are suitable), nor does it help.

 

Now that makes sense. And combined with your original post, I think provides a logical explanation for our previous "sub-optimal" results. The MM trainer we worked with told us it was okay to calibrate if he was out of range, as long as he was stable.

What you described is pretty much what we’ve been doing the past couple of weeks, I check his BG and calibrate before I wake him up, because he’s usually pretty stable at that time of day (and now I’ll also be watching for him to be in range). I’ve told him not to calibrate at school. If he’s stable when he gets home I’ll calibrate before his afternoon snack, if not I’ll look for another opportunity before dinner. Then if he’s stable at bedtime, I’ll calibrate a third time. If not, I’ll stay up till he is stable and calibrate then.

I think now I’ll be less reluctant to do additional calibrations as opportunities arise, because sometimes the calibration clock runs out on me (without a stable time to get a BG), and I’m forced to either use a BG that I’d rather not or stop getting readings.

Any idea how many prior readings are factored into the algorithm? In other words, how many good calibrations does it take to overcome a bad one?

 

Even if BG is stable, calibrating out of range is not optimal because the CGM's sensor uses an enzymatic reaction
that is not linear over a wide range of BG. If you have no choice to cal when BG is stable but out of range, it's good
to cal again as soon as BG is stable and back in-range.
​

Any time you can get an in-range stable BG is a good time to cal. Of course, those times will be different for each
child, and may even vary depending on the time of year.
​

I can't know this for sure... certainly MM keeps it proprietary. However, I've always had great accuracy after any single
good calibration. Even if prior calibrations are taken into account, accuracy can be restored with one good cal.
​

The only time that 2 or 3 cals are usually needed are for a new sensor. This is not because of the cal algorithm, but
because it can take the sensor 6 to 12 hours to fully stablize from a biological perspective. When you re-start the same
sensor after 3 days, the sensor is well stabilized and the first good cal will give you good accuracy.
​