Pretty sure the -70 is an artifact of the fact that Dexcom is having trouble keeping up with trends. But do others think it’s really possible to drop 70 points in 5 minutes?
1 Like
Yes. I’ve seen this as previously indicated in another topic…actually done finger prick checks and verified +100 point drops (or rises) in a single tick. It’s not something that happens frequently, but have seen first hand that it does happen (and not just a CGM unable to keep up thing).
1 Like
Mon son has gone up (for real, on a blood glucose meter), more than 100 points in 10 minutes – it is almost routine for his hormone peaks. We have seen several times, in the middle of a big correction, his Dexcom BG go down by 45 between two points.
My son weighs about 120 lbs, so I figure that with a much lighter patient like yours, reaching 70 could make sense.
1 Like
If it isn’t an artifact of data problems (which I still suspect it is, as the following data point is almost always just -10 or -15 from the last one), then I suspect it must be an insulin pooling problem, where the insulin is being delivered into the fatty tissue but not migrating into the blood stream, and then a big glob of it all hits the bloodstream at once. There is no other explanation in my mind. Because he still has a belly full of food on board, and no other hormonal things we were dealing with prior.
I do agree blood sugar can sometimes jump 30 or 45 points in 5 minutes (After all, the liver has a lot of sugar it can quickly dump if it gets the signal), but in clamp tests they’ve shown that at least with adult T1Ds, it’s pretty impossible to spike more than 400 -500 points in an hour, which isn’t a sustained rise of +100 per 5 minute period.
Also, we already know that Dexcom is measuring interstitial fluid, which is 15 minutes behind blood at some points in time. It’s using an algorithm to catch up. So it seems like a much more reasonable assumption that these rapid changes are a sign that Dexcom’s algorithm was not keeping up in the 10 or 15 minutes prior to the crash, rather than an authentic drop of 70 points in 5 minutes. For our son, to drop 70 points would mean 0.3 units of insulin would all need to be active at the same 5 minute-window in time. Today he only got about 1 unit for his meal of about 40g of carbs, and if his DIA is 4 or 5 hours, that basically means he got a whole hour or even 1.5 hour’s worth of insulin all acting at once – and that his food wasn’t in any way causing a blood sugar rise at the same time.
I guess I am more likely to see very sharp rises as a true phenomenon, especially after a food underbolus, but these random drops that are then followed by a very normal decrease or even plateau seem more likely to be an artifact of Dexcom’s failure to keep up. In general I think it overestimates lows, is slow to catch up to them, and then overcorrects up once they start rising. I can see at the other end of the curve (fast spike, turning the corner, then starting to drop) it might do the same thing.
But I guess the only way to know for sure is to test his finger – and he’s at daycare where they don’t often do that, so we’ll never know on this one 
1 Like
@TiaG, many insulin Dexcom drop curves are nice and regular for us. But, occasionally, some start slowly, then do almost all their drop in 3 points, then end slowly also. It might go -3, -3, -3, -12, -40, -15, -5, -3, -1. I have not found any categorization to explain them yet. I don’t think it is an artifact, but I can’t prove it.
The real hormonal up-curves, on the other hand, are often sudden and violent. After a slow start for 5-10 minutes (you don’t really know if they will take off), they reach their real (finger prick) max almost instantly, within 10 minutes, then stay there until insulin has caught up. The Dexcom, interestingly, after climbing regularly (that I am sure is due to the algorithm, because it seems to drop faster) sometimes takes a long time to get the last few points before it fully catches up: it can be “almost there” for 20-30 minutes sometimes, a weird phenomenon.
OK, I found a drop that is exactly the same as yours, from last Saturday:
I don’t think it is an artefact, because it went up and down a while right before.
I don’t have manual BGs for this stretch, but, when you look at the peak going up, at 1029 it is 199, at 1059 it is 296. I can’t prove it but I think this is likely a 100 point rise in 10 real minutes – the Dexcom did not catch up for a while.
It is a lot harder to find a documented real BG rise in our logs because I don’t have plots for them: I have to scroll page by page for every day, which takes a long time. I may or may not have the heart (or the time) for it 
As a note: I find that the Dexcom typically undercounts the high peaks by quite a margin, because it misses them. I will see many peaks on a fingerstick that the Dexcom never reaches by 30-40 points.
[EDIT] Improved precision of left-side arrow positions