Looping and extremely frustrated with settings; maybe someone here has suggestions?

[TiaG]

For the last several months now, Samson’s blood sugar has been much more erratic. He’s been having lots of highs and lots of lows – often the highs are at very consistent times of day, so that makes me think they’re fixable. The trouble is I’ve been trying to fix his settings because I know for sure they’re wrong. But the problem is I get contradictory results base on basal testing versus other patterns.

The standard advice, in everything from John Walsh’s Pumping Insulin to the Looped group, is to check basal settings with a fasting BG when all the bolus insulin is out of your system. Luckily, since Samson is a picky eater who gets very involved with his toys, he often isn’t ready to eat “on time”, and we have a few hours here or there where we’ve been able to do basal testing without him even realizing we’re doing this test. Plus, one day last week he was so tired he just came home at 4:30, fell asleep and slept all the way till 6am the next day. So we got a good long stretch from about 9pm to 5am where he had no bolus insulin in his system, his settings were just left alone, and I had to give him low treatments throughout the night. I eventually even gave him a 0 temp basal for a few hours to get him up.

So when I do the fasting basal testing, I am seeing several drops in blood sugar, as if the basal is too high. BUT, on a day to day basis he usually runs high for several hours (11am to about 4pm, then again from about 7pm or 8pm to midnight), and often crashes low several hours later, which suggests an overly strong mealtime bolus to compensate for insufficient basal. When I run autotune, it suggests my ISF is too strong and my basal is too weak – essentially recommending the opposite of what the basal test is recommending.

My ISF is probably too strong because I see a lot of those hills and valleys where Loop is mandating a high temp basal very quickly, and then it shuts off abruptly as the blood sugar begins to nosedive. Given that we’re also spending a lot of time high, that would suggest the basal is too low. This is the direction my endo is also leaning.

But, i guess I’m wondering if all this confusion is because food somehow takes a lot longer to be digested for Samson, or at least that it affects his blood sugar a lot longer than all the absorption models assume.

I’m feeling a little paralyzed and not sure which way to go next. Samson is starting a new school next year and I want to have his settings dialed in and reasonably close to ideal by September. I’m just not sure what to change first. We have about 10 weeks to get things dialed in.

[bkh]

@TiaG

So we got a good long stretch from about 9pm to 5am where he had no bolus insulin in his system, his settings were just left alone, and I had to give him low treatments throughout the night. I eventually even gave him a 0 temp basal for a few hours to get him up.

That does sound like too much basal. So reduce it a little starting about 1.5 or 2 hours before it needs to take effect, and that should help.

@TiaG

My ISF is probably too strong because I see a lot of those hills and valleys where Loop is mandating a high temp basal very quickly, and then it shuts off abruptly as the blood sugar begins to nosedive.

Ok, so that’s one thing to start fixing. Recently I was seeing those kinds of waves in the CGM graph and I gradually backed off the ISF from 38 to 46 before the oscillations damped down.

@TiaG

But, i guess I’m wondering if all this confusion is because food somehow takes a lot longer to be digested for Samson, or at least that it affects his blood sugar a lot longer than all the absorption models assume.

You could explore that by telling loop a longer digestion time for the meals, like if you typically use 3 hours try 3.5 or 4 and see if that helps. When my mealtime graphs are showing an early rise followed by a low later (assuming I think my basal is right) then I start thinking I needed a smaller meal bolus but pre-bolused earlier. I don’t know if it’s feasible for Samson, but if you announce a meal to loop in advance, it can do a much better job. Like at 3:46 you can enter 38g carb, 3 hour digest, to be eaten starting at 4:30, and loop will start pumping in the pre-bolus to get a good amount of active insulin in the bloodstream by the time the eating starts, and that really can help blunt the meal spike.

@TiaG

I’m feeling a little paralyzed and not sure which way to go next.

You’ve already identified ISF as one thing to work on, and the meal bolus as another. So start making small adjustments and see things begin to work better.

Just in case you haven’t found it yet, https://kdisimone.github.io/looptips/ has some concepts and ideas that may help.

[TiaG]

Thanks @bkh

That does sound like too much basal. So reduce it a little starting about 1.5 or 2 hours before it needs to take effect, and that should help.

so yes! I agree. But the odd thing is that if he, says, eats dinner, this pattern is totally different ALL NIGHT LONG ,like until 5 AM. Here’s a picture of a night when he ate a fairly substantial dinner of carbs:

And compared to the night when he fell asleep with no dinner:

Even though there are only a few blips into the “red” low zone, that is hiding the fact that I was feeding him glucose roughly every few hours all through the night.

That’s what makes me confused… it really does seem to me, based on the patterns that we see, that his meals affect how much insulin he needs for literally 7 or 8 hours after, not just 3, or 4 or even 5. We rarely prebolus and most of the time he doesn’t spike that much – and he’s flat or goes low after he eats and then he spikes usually 1.5 to 3.5 hours later. And we already have tried things like logging pizza as 8 hour-absorption, logging extra carbs later on in the night that add up all the fat and protein, etc… and it isn’t enough to keep him in range usually, unless we just completely make up a crazy number of carbs. So I guess that’s what is always hard. I have this hunch that something is up with Samson’s digestion and it really takes a long time for food to leave his system, but if his food always takes like 8 hours to leave his system, then does it really make sense to have his basal set as what it would be when he hasn’t had food in his system for hours?

We are trying to weaken the ISF for now, so I raised it from 250 to 260 and will keep going. Autotune is recommending 290, so I may still have a ways to go.I 'm pretty confident what we have is just too strong. If we wind up seeing persistent highs after those changes, that’s when I’ll tweak the basal or the carbF I think.

It’s just so tricky.

[T1Allison]

@TiaG, as you know I do not loop and I was not T1D as a child…but I thought maybe my own observations with spikes/tanks might be useful in some way to you guys.

1. My basal settings keep me flat as long as I’m not eating. If I’m eating, I actually need more basal. Depending on the circumstances, I’ll either add +10% basal during eating/digesting hours or I’ll just add to meal boluses and be prepared to catch the downturn if it ever happens.
2. My bolus control is super tricky for me…and what I mean by that is this:
   a. Prebolus timing is everything…but the amount of prebolus time I need changes based on my activity level which makes it perpetually tricky.
   b. If I just add to my bolus to make up for the fact that I don’t have the time (or confidence) to prebolus appropriately, what tends to happen is that I’ll meal spike up to low 200’s and be stuck there for 3 hours…and then Novolog will finish strong and tank me later. My boluses do all eventually work…just not when I want them to when I’m trying to apply my own wishful rules to them…which is frequently. What I mean is that if I overbolus to make up for timing issues, I will pay for it later.
   c. Unless I’m being non-stop active during the day, I cannot eat all of my meal carbs (that I’m bolusing for) all at one time. I’ll have to stretch them out over two hours to stay respectably close to range (for me…I’m just hoping to keep my meal spikes under 180). If I do eat them all in one shot like I did before I was T1D, then I’ll spike, linger, and eventually come down five hours later. And if I overbolus then I’ll get a tank at the end.

Good luck! Everyday is different for me and I can only imagine how it is with a little one.

[elver]

@T1Allison

My basal settings keep me flat as long as I’m not eating. If I’m eating, I actually need more basal. Depending on the circumstances, I’ll either add +10% basal during eating/digesting hours or I’ll just add to meal boluses and be prepared to catch the downturn if it ever happens.

Wait…what!!! I’ve never heard or even thought about that strategy! I have delayed digestion issues that that strategy would be perfect for. Thank you so so much @T1Allison for the great idea.

[T1Allison]

@elver

Thank you so so much @T1Allison for the great idea.

Good luck!

[bkh]

@TiaG

That’s what makes me confused… it really does seem to me, based on the patterns that we see, that his meals affect how much insulin he needs for literally 7 or 8 hours after, not just 3, or 4 or even 5.

I can think of two possible explanations. One possibility is unusually slow digestion. If that’s it, then loop needs to be told the extra-long absorption time as you’ve already tried with pizza. But I can think of a second possible cause. I know that in myself, if I don’t get enough insulin early enough (i.e., a pre-bolus sufficiently early before eating) then my BG rises more than I want. And if my BG rises over 160 or 180 and stays there for a little while, that causes me to become significantly insulin resistant, so I’ll tend to stay high despite repeated large corrections until I finally go far enough to cause a crash. I don’t know the scale on the BG graphs that you showed, but if the first example is a rise up to 180 or so (from 5:30pm to 6:30 or so) then maybe there’s some extra insulin resistance here, and maybe you’ve been treating that with extra basal. To see if that’s it, try to prevent the initial spike, and see if that is enough that you don’t need extra basal to keep things under control.

@TiaG

It’s just so tricky.

So true. If it were only a matter of counting carbs and doing the arithmetic with a correct insulin:carb ratio… Fortunately for Samson it is abundantly clear that you have the will and the smarts to cope with the actual complexity.

[MM2]

@bkh

And if my BG rises over 160 or 180 and stays there for a little while, that causes me to become significantly insulin resistant, so I’ll tend to stay high despite repeated large corrections until I finally go far enough to cause a crash.

I have same experience, not using loop. Stacking insulin gets me a treat at some point !!

[elver]

I used your technique tonight with great success!! I am hoping to replicate this result in future days/meals. Thanks again @T1Allison for the great idea

[TiaG]

I mean the thing is, he doesn’t necessarily spike super high immediately after a meal. What often happens is he is bolused, spikes up briefly, comes down to nearly in range and then an hour or two later spikes up really high. (Although the particular example I posted, I don’t remember exactly what happened but he was clearly on the rise before we bolused him.) We used to prebolus in the beginning with Samson and we stopped because, while it would leave him very flat initially, he would always have a rise about 1.5 to 2 hours later that was almost the same magnitude as if we didn’t prebolus. It felt like his meal outlasted his insulin. And then if he got picky or ate less than we expected, we’d have the unpleasant task of feeding him tons of glucose tabs.

We definitely DO see that insulin resistance that requires a ton of boluses to come down, only to lead to a crash eventually. But the weird part is that it’s not usually the first spike after a meal that needs this; it’s that second spike.

[TiaG]

1. My basal settings keep me flat as long as I’m not eating. If I’m eating, I actually need more basal. Depending on the circumstances, I’ll either add +10% basal during eating/digesting hours or I’ll just add to meal boluses and be prepared to catch the downturn if it ever happens.

This seems like something that definitely happens with Samson. Not sure how to accomplish a +10% for Loop, but maybe I can set up an override preset of +X%.

c. Unless I’m being non-stop active during the day, I cannot eat all of my meal carbs (that I’m bolusing for) all at one time. I’ll have to stretch them out over two hours to stay respectably close to range (for me…I’m just hoping to keep my meal spikes under 180). If I do eat them all in one shot like I did before I was T1D, then I’ll spike, linger, and eventually come down five hours later. And if I overbolus then I’ll get a tank at the end.

For Samson, it’s almost the opposite. Other than for breakfast, when we just bolus for all the carbs no matter what, he can eat 100 grams of carbs at a meal, and we still will only bolus for something like 40 to 50 grams. Because if we bolus for all of them, he’ll go super low for a few hours, then spike high for 7 or 8 hours as the rest of the meal digests (this is especially true for dinner). But I definitely feel like him eating carbs all at one sitting is a PITA because it is harder to manage. I do agree the tail phenomenon is a big factor for us, and we have a lot of those 5-6 hour tanks. Usually that’s after stacking for sure.

[bewell]

@TiaG

I mean the thing is, he doesn’t necessarily spike super high immediately after a meal. What often happens is he is bolused, spikes up briefly, comes down to nearly in range and then an hour or two later spikes up really high.

You may wish to consider whether the meals after which he encounters the delayed spike contain high protein or fat. The following explanation is from our diabetes educator (slightly edited to remover personal references)

"You might be able to find something on the “Protein-fat-Unit” which was conceived in Poland I believe. Essentially, it is an estimation of the amount and duration of insulin action required for elevated glucose following high fat/protein containing meals. It goes something like this: treat each 100 Cal of fat/protein (in high fat/protein meals) like 10 grams of carbohydrate, but extend the 1st 100 Cal over 3 hours, and every additional 100 Cal over another hour. So 30 grams, as an example, could be 250 Cal (fat is 9 Cal/gram, protein 4 Cal/gram), or 2 - 2.5 PFU’s, which according to your insulin/carb ration works out to about 2-3 units of insulin delivered over 3-4 ½ hours. That would be in addition to whatever insulin is needed up front for carbs at the meal.

Another way to do it is to simply add 30% more insulin to your meal calculation for carb intake (when having something high fat/protein) and extend it 3-4 hours. It obviously takes some fine-tuning, but it does help deal with the prolonged elevation in blood sugars following certain meals."

[Trying]

@TiaG

I mean the thing is, he doesn’t necessarily spike super high immediately after a meal. What often happens is he is bolused, spikes up briefly, comes down to nearly in range and then an hour or two later spikes up really high.

This happens to me, too, primarily only for dinner, which could be as @bewell says, due to unaccounted protein or fat. Pre-loop, I would just always take an addl bolus of 1U, one to two hours after eating, even if my BG was in range at the time, on the first slight uptick in BG. If I still go up an hour later, I’ll take another 1U. That usually prevents any later spikes and keeps me in range, but sometimes can lead to a low, so small carb may be needed. These extra units could be required due to any fat or protein in my meal, I’m not sure.

Now that I’m Looping, I either let let Loop perform the increased temp basals, or enter fake carbs to justify correction boluses of 1U as described above during for pre-Loop.

Another issue this could be related to is simply being insulin resistant due to OmniPod pod change on day 1. I experienced this today I changed my pod this morning at 6am. Pre-bolused 2U for breakfast of 30U which normally works well. Today I had to continually give bolus after bolus, a total of 5U now, plus Loop performed 4U/hr (scheduled is .5U/hr), for an hour! I probably should have just given myself a syringe bolus, but I read that giving the extra boluses from the pod should help remove any bubbles in the pod. Last pod change that technique worked, but so far, not today

[bkh]

@TiaG

I mean the thing is, he doesn’t necessarily spike super high immediately after a meal. What often happens is he is bolused, spikes up briefly, comes down to nearly in range and then an hour or two later spikes up really high.

That sounds like what happens when the ISF is set too strong. Loop sees the initial spike from the meal, overreacts for a little while, then when the BG quickly turns down loop panics in the opposite direction and suspends the insulin delivery, and that leads to a second spike. Weakening the ISF will cause loop to react more gently rather than pushing too hard down and too hard up. It will be interesting to see if Samson’s double spikes get easier to handle when you get the ISF and other knobs dialed in better.

If it’s still an issue, then I’d try entering the meal in two parts. One smaller portion pre-bolused and with a fast clearing time like 1.5 hours or 2 hours to solve the initial spike, followed later by a second portion with a slow time like 4 or 6 or 8 hours. Two spikes, two treatments.

@bewell

You may wish to consider whether the meals after which he encounters the delayed spike contain high protein or fat.

Good idea. I heard of an approach called TAG bolusing (TAG=total available glucose) that’s supposed to handle the fact that a portion of the fat is converted to glucose over a period of many hours, and a significant portion of the protein is also digested to glucose, faster than the fat but significantly slower than the carbs. There are lots of variations, but one simple one was to bolus for the carbs normally, and also do an extended bolus (like 6 hours) for 40% of the protein grams and 10% of the fat grams. This may require weakening the carb ratio, which previously only counted carbs but was expected to cover the protein and fat too.

[Beacher]

@bkh

one simple one was to bolus for the carbs normally, and also do an extended bolus (like 6 hours) for 40% of the protein grams and 10% of the fat grams. This may require weakening the carb ratio, which previously only counted carbs but was expected to cover the protein and fat too.

My experience from the days when I counted protein (30%) and fat (10%) was definitely to scale back on the carb ratio. I had a really good A1c then, but I was perpetually treating lows. It just wasn’t worth it. Now I use extended boluses or increased temp basals for high-fat or high-protein meals.

[TiaG]

I feel like this numeric approach sounds good… when we are more on top of carb-counting we may try it. We did go on one of the hot days to an Olive Garden (which is sort of a ridiculous place – their menu and calorie-counts gave me hives). But, it has ALL the nutritional content so we bolused Samson based on the protein and fat in the kid’s cheese pizza and it worked great that one day. Just one data point, but it was encouraging.

I think we need to be a little more proactive with the actual carb-counting too. We’ve been SWAGGING things for too long and sometimes it’s unavoidable. (homemade baked goods, three pieces of pizza with 2/3s of them nibbled erratically away). But we definitely have gotten sloppy on measuring/bolusing for other stuff where we theoretically could, like a bowl of pasta or a slice of bread, etc… I guess we’ve just been in the “reactive” mode for a while so the benefits of carb-counting meticulously felt obscure… we needed the other settings to be right before we tweaked those, I think.

[TiaG]

That sounds like what happens when the ISF is set too strong. Loop sees the initial spike from the meal, overreacts for a little while, then when the BG quickly turns down loop panics in the opposite direction and suspends the insulin delivery, and that leads to a second spike. Weakening the ISF will cause loop to react more gently rather than pushing too hard down and too hard up. It will be interesting to see if Samson’s double spikes get easier to handle when you get the ISF and other knobs dialed in better.

YES! I did weaken all the ISF’s by 10 points, but I suspect we still have a ways to go. I’ll update this thread to see how his settings change over time. Hopefully this will keep me motivated to keep on this, rather than getting distracted by life.

For reference, these are his settings right now:
6/17
Carb ratios (grams carb/1 unit insulin):
00:00 a.m. | 30
05:00 a.m. | 20
17:00 p.m. | 25

ISF (points/unit of insulin):

05:00 a.m. | 260
16:00 p.m. | 240
23:00 p.m. | 330

Basal rates (units/hr):
12:00 a.m. | 0.125
3:00 a.m. | 0.2
4:00 a.m. | 0.225
6:00 a.m. | 0.25
8:00 a.m. | 0.2
12:00 p.m. | 0.4
2:00 p.m. | 0.225
7:00 p.m. | 0.275
9:00 p.m. | 0.4
10:00 p.m. | 0.425
11:00 p.m. | 0.3

[TiaG]

So here was what yesterday looked like with weakened ISF:

Yesterday was flatter than usual, so that’s good, but there was one pretty bad crash in the afternoon where Samson started rising and then Loop piled on a bunch of insulin, so I am weakening the ISF by 10 points again.
Then in the middle of the night, he had no Loop running (he snuck into our bed out of range of RileyLink), and he was just on his programmed basal settings and slowly rose from about 110 to about 200 mg/DL over two hours and then was just stuck there. I didn’t bolus him because I wanted to see what would happen on its own. When he went back in range of his RileyLink, Loop piled on a bunch of insulin and led to an early morning drop and low. So I think that the basal rates in the midnight range could potentially be a little low (although the rise happened about 2.5 hours after his meal, so possibly not), but the ISF is definitely still too high. So for now I’m changing the ISF another 10 points.

6/18 settings

ISF (points/units of insulin):

05:00 a.m. | 270
16:00 p.m. | 250
23:00 p.m. | 340

We’ll see how that works today!

[bkh]

Nice results. Your analysis is good and the results are heading in the right direction. Keep up the good work!

[Katers87]

I know that this thread is for your benefit

But I’m really enjoying learning about loop, and those graphs are awesome. It’s great that you can see how loop is responding so easily and make small adjustments like you are. Everything I’ve read about the 670g seems like it’s this black box that is really difficult to manage. Loop looks like a significantly better tool.

Thanks for sharing your troubleshooting strategies. His numbers yesterday look great for his age! Not that they couldn’t be better with additional adjustments to his ISF (I have little experience with this, so I couldn’t say).