Transient peaks after glucose intake?

Last night we got trapped again by dosing a transient peak that we sometimes see after glucose intake.

My son had slightly overcompensated for a low after dinner. His CGM was at 139. He fingerpoked and got a 156, then a 170. He decided he needed to dose, and corrected with 2.5 units of insulin.

I was a bit worried, as I remembered that after a correction we often see a real (i.e. measurable by a hand glucose meter) transient peak that never seems to show on the CGM. Sure enough, the CGM never went above 139, and the correction was way too strong. If he had dosed for 139 he would have been just right.

Do you often see this transient peak? How do you deal with it? Why do you think we don’t see it in the CGM? What do you think causes it?

This is not the first time we have gotten trapped that way. Btw, his calibration was good, within 5 of CGM, so the discrepancy is not due to a calibration issue.

1 Like

Fast transient peaks will always be a difficult CGM issue. The reasons are somewhat nuanced, but think of it this way.

In order for the CGM to pick up a bg measurement, the blood sugar needs to come into an equilibrium with the tissue glucose, then at the sensor tissue interface you have to set up a second equilibrium because every sensor has to have a way to keep most of the glucose away from the sensor, otherwise you would overwhelm the sensor with signal. So to get a reading, you need to have two processes come into equilibrium before the sensor will be able to “see” the glucose change. Fast transient will never be great with the current sensing paradigm.


I think about it as the consequence of diffusion of glucose from the bloodstream to the interstitial fluid until it reaches the sensor, and then the sensor does some modest additional smoothing to remove noise. So a step-change in BG will correctly reach the CGM (after a delay), but a brief spike will be averaged out by the diffusion process: the peak becomes lower and broader. I think about it as analogous to dispersion in an optical pulse traveling through a fiber, although the processes are different.


If “after glucose intake” involves handling glucose tabs, also make sure that you’re not picking up some from the skin during the finger poke. That’s one of the scenarios where I make sure I wash my hands first.


Yes, frequent problem, we are very thorough on that. He washed his hands again before the second fingerpoke to make sure.

The other thing we do is that we use tubes for glucose that he can feed from, so in general he doesn’t have to handle sugar with his fingers. We use Skittles in glucose tab tubes that slide easily and can be counted with your tongue:-)

1 Like

Are you asking about the CGM discrepancy, or the quick BG spike that came back down?

1 Like

About the quick BG spike :slight_smile:

Insulin acts on the glucose in the blood. Glycogen synthase activity increases, and this allows the cells, liver, and muscles to take the glucose and use it. Glycogenesis starts, it is the conversion of glucose into glycogen.

Once the glucose is in the blood, insulin is able to start the process of glycogenesis. If the glucose was not in the blood, none of that would happen.

If you eat fast-acting sugars, you see a quick spike, just like a non-diabetic before their insulin has responded. The non-diabetic’s spikes are shorter because their insulin is delivered directly into the bloodstream.

Not exactly sure what you mean by transient peak: very sharp increase after some fast carbs (sure, nothing unusual about that), or very sharp increase followed by very sharp decrease, none of which are captured by CGM? If the latter, I’d consider that unusual. The only situations when I’d see bg dropping at a rate comparable in magnitude to the up-tick due to fast carbs would be due to exercise or a (pretty substantial) over-bolus or bolus correction. Otherwise, I’d suspect noisy meter readings. What happened after that 170 reading? If in the next few minutes the meter reading was well below 170, than I’d say almost certainly meter noise (especially given that he washed his hands). The first one (159) - closer to CGM but slightly higher - was probably closer to real bg.

1 Like

I do see the peaks on the meter and finger stick. Remember, a bolus of insulin will probably take about 15 min. to take affect. Meanwhile your bg will continue to do what it was doing. I would wait another 15 min to half hour, and then finger poke again. I bought a book, “Think Like A Pancreas” (which I may have learned about on this site), and it has helped me with some issues. It may be available at your public library.

1 Like

This is a very non scientific explanation, but I feel like we often observe a quick peak after a low treatment (or eating a fast acting carb like a low fat, mostly sugar dessert without a long enough pre bolus, which is sorta the same thing) which under some circumstances, “flops” shortly thereafter. It seems like those circumstances are when we have a lot of insulin on board relative to “real food” on board, or when he’s glycogen depleted (like post exercise lows). I think of it as titration – the sugar is absorbed via the gut and appears in the bloodstream all at once, the insulin acts on it, and it is sucked back into the cells. If there’s little or no buffer present in the system-- slower acting carbs and other foods continuing to ooze out into the bloodstream, or glycogen stores being released – that peak goes “flop”. Our experience has been that if we manage these particular peaks with aggressive bolusing, we get burned by the “flop” setting up a cycle of chasing highs, chasing lows. So I often wait it out for an hour. But it’s not always predictable which peaks will go “flop” and this may just be the flip side of you guys aiming for such nice tight control–we may not be doing as well there.

What others have said about the smoothing of the Dexcom curve, relative to what you can observe with frequent fingerstick, certainly makes sense to me. I guess the question is whether you’re better off acting on the more minute to minute fingerstick data, or if management decisions based on the smoothed and delayed dexcom trend would give decent or even better dosing decisions on these quickie peaks. I say that because you’re really concerned about trend – you’re not deciding to correct because your BG is 170 this instant and that’s so awful, but because you’re taking that 170 as a sign that you’re headed up and you want to reverse that trend. In other circumstances (like, oops we under dosed that high fat meal, or forgot to turn basal up before vegging in front of a movie and now we are on a rocket to the moon) I WOULD trust the finger stick over the Dexcom and bolus full correction because my intuition is that’s not a high that’s going to “flop” but just continue on up.

Again full disclosure that I am not claiming flop prediction is based on real science!! :wink:


There’s another CGM effect that can make a spurious quick spike on the CGM that doesn’t reflect actual BG. It happens because the reading the CGM shows is not a measurement, it is a prediction. The changes in glucose concentration in interstitial fluid lag the capillary BG by 5 or 10 minutes, so the CGM algorithm basically is predicting 5 to 10 minutes into the future based on its actual measurements, and then showing the predicted value as “current” BG. This works well when the trend is straight, but at a turning point (like when rocketing up and then sharply turning back down) they’ll be predicting the highs as if the actual BG rise continued for another 5 or 10 minutes. So that mechanism causes an erroneous display of higher highs around the reversal time of fast moves higher (and lower lows when falling fast.)