I had a sense that reverse corrections were not working for us and now am very concerned about how our PDM is calculating them. In many cases, it suggests a bolus that I believe is .05 u. lower than it should be -- in all cases when there is no IOB. I noticed this late Saturday afternoon when DD's BG was 85 and she had a 12 g. snack and no IOB. At that time, her Target is 120, I:C ratio 1:50, and ISF 1:200. The PDM suggested 0.00 u. correction. By my calculation, she should have received a 0.05 u bolus: Reverse correction of -0.15 (rounded down to the nearest .05 u) BG - Target / ISF = Reverse Correction (85 - 100) / 200 - -.18, rounded down to -.15 Carb Bolus of 0.24 12/50 = 0.24, rounded down to .20 Total Bolus -.15 + .20 = .05 u In checking the CoPilot software, I see this happening nine times in the last 10 days or so. In each case there has been no IOB (it's been easy to check as I check DD's BG on waking and bolus her for 8 g. gummi vitamins). I called Insulet who gave me the name of a CDE in my area who is supposed to be able to explain this, but I am going to wait until I also have DD's PDM in hand to double check the calculations shown there against the Co-Pilot download (though the example from Saturday was one I noticed and refigured using the PDM). Given DD's low insulin needs, 0.50 u [ETA: should say 0.05]is a meaningful difference. Now I'm wondering if there could be an issue with our PDM, this is not complicated math, so I don't understand how the PDM would be doing this calculation differently.

I don't know how Omnipod does their rounding, but I would round -0.18 to -0.20, and then I would also get a 0 bolus. There are two ways to look at it... -0.20 is the "nearest" number to round to. Or -0.20 is the "safest" number to round to (ie the pump algorithm always decides to give less insulin in a rounding situation). You may have other examples that look off, but this one looks okay to me.

as a side note, have you checked the ISF lately? now that you are pumping? To me at least it seems like that is a fairly strong correction factor for someone with a 1:50 ratio. So after these many times of underbolusing, is she ending up above the target120?

I think you are right that it must have something to do with how it does the rounding. And I agree that -0.18 ought to be rounded to -0.20, but in that case I would think that the 0.24 u for the carbs would be rounded to 0.25, which would give you a 0.50 u [ETA: this should say 0.05 u] bolus. So now I'm wondering if all reverse corrections are rounded up and all boluses rounded down -- I'll have to look at the numbers again.

Well, rounding -0.18 to -0.20 actually is rounding down since it is a negative number. At least, that is how I look at it!

0.50 or 0.05? You've said both and i'm not sure which you are meaning. I agree that it sounds like its more of a rounding situation if its 0.05 and I wouldn't be worrying about 0.05.

We started using 1:200 (at the suggestion of our CDE) when we began pumping. It has seemed to be work pretty predictably and corresponds to some of the formulae for determining ISF. Yes, prior to this, I had a general sense that reverse corrections did not seem to be "working" and when I saw the suggested zero bolus on Saturday, it led me to look at the numbers more closely. Now I'm just scratching my head.

Sorry. I did mean 0.05. I understand that for others, that might not be a concern. The thing is, with DD, 0.05 u -- whether a missed basal pulse or a bolus for carbs -- seems to really make a difference. When we started pumping the CDE characterized DD as being "sensitive" to insulin; I don't know if that is because she is still honeymooning or not, but the dosage seems to make a difference, which is why I'm puzzled by how the numbers are being rounded.

I come up with +0.07U without rounding - +0.10 when rounded. EDIT: I don't think it's legit to round until all the math is over.

I agree, but the PDM does. You can display the calculation used to arrive at the suggested bolus and the individual elements are rounded. Unfortunately, that data does not download into the CoPilot software, so I have to look at that separately when DD gets home.

I thought about this more ... and I was wrong about the rounding, I think. It's rounding to the nearest 0.05 which, in this case would be 0.05 (7 is closer to 5 than 10). EDIT: I think it's generally accepted that -0.18 would round to -0.20 mathematically. If it is rounding each step then that would explain the 0.00U, right?

You can always override the PDM if you don't like what it's come up with. It makes sense to me that it would round to give less insulin when under target. We rarely use the reverse correction, though. I'd be perfectly happy with a pre-meal 85.

I think that this isn't really a pod malfunction; instead it is a standard way that the pod works and your ISF needs to take it into account. Out of curiosity - in the 9 out of 10 times that the pod missed 1/500th of a unit, how did it affect her numbers afterwards? We are talking a reverse correction - so did she go really high? Or just not high enough? If her sensitivity factor is 1/200 - 1 unit of insulin will lower her 200 points (or 1 missed unit will raise her 200 points) - then I guess I am scratching my head why 1/500th of a unit keeps causing you so many problems? This should only affect your BS by 10 points max if your sensitivity is correct.

I agree that it is not a pod malfunction -- but I need to understand the "standard way that the pod works." So far, I have not been able to, despite me and DH looking at the numbers, figure out what the standard way is. In most cases -- but not in all cases -- it appears that the reverse correction number is rounded up and the carb bolus number is rounded down. I have looked at the numbers with standard rounding, rounding all up, and rounding all down, and cannot find the "standard way that the pod works." I have a call into a clinical services manager for Insulet. I had -- as a general matter -- noticed, particularly going in to dinner, that DD's numbers were more predictable when she went into dinner at or above her target and that it seemed that whenever we did a "reverse correction", we got much higher post-meal numbers and wound up doing nighttime corrections. We started using the 1:200 ISF at pump start in the hospital. I agree that, ideally, .05 should only involve 10 points. But it does not seem to. And the reason that I particularly noticed it was that the PDM's suggested bolus for the 12 g. snack was zero. Based on my experience, obviously limited to my DD and only since last November, a 12 g. snack with no insulin would be a problem (we were doing uncovered school snacks on MDI). It would not surprise me at all if her ISF may need to change -- the 1:200 roughly correlates to some of the formulae out there and seems to work -- that may be because most of our big corrections are in the first half of the night when she seems to be a bit insulin resistant. We are continuing to work on DD's nighttime basals and I think once those are improved, we may well end up changing the ISF. I get that others may think that I'm making too much of this. That's fine. We are dealing with a multitude of variables. But how a suggested bolus is determined when a reverse correction is involved (and there is no IOB) should not be one of them -- it's math, not particularly complicated math. For the moment, I'll just recheck all the PDM's calculations and override the bolus as needed.

I agree with the pre-meal 85, however, as DD's bedtime is 7:45, all other things being equal (and assuming this would even remotely be in my control ), I'd prefer a bit of reverse correction when the pre-meal 85 is at dinnertime. Once I understand how the PDM is actually doing the calculation, I'll be more comfortable overriding the suggested bolus (we do that now to address IOB for a meal/snack bolus), for now, I'm just redoing the math myself.

Well, the carb bolus would have been .24 and the reverse correction would have been -0.18. So it looks like it rounded the reverse correction up to -0.20 and the carb bolus down to .20. Of the 10 or so examples I found in the last couple of weeks, that approach accounts for all but one of them (where reverse correction would have been -0.205 and carb bolus 0.400 and suggested bolus was 0.20, so it looks like the reverse correction was not rounded). I'm thinking that there must be some fairly standard rule that the PDM is programmed to apply -- it's frustrating to be unable to figure it out (I've not found one single approach to rounding that explains all of the examples).

We simply don't use the reverse correction feature on the MM, because I don't like how it calculates it. I shave off a few carbs when he goes into a meal low, and do not enter a BG.

We did not use reverse correction on our old cozmo and do not use it now with Omnipod either. I know it should work in theory, but it just doesn't for us. I can give ds a couple fruit snacks if he's lower than I want him to be or just shave off a few carbs like Hawkeyegirl said.

That's what I do, too. I know, in theory, the reverse correction is supposed to work, but that's not the case for us.

I spoke with an Omnipod representative today. She confirmed what one of the PP's surmised -- that the "suggested bolus" is determined by "rounding" the calculated carb bolus down and any correction down minus any IOB (based on the pump's linear formula, which is in the manual), so that the "suggested bolus" will be the lower, more conservative amount of insulin. So if the math results in a carb bolus of .24 u, the PDM will round that down to .20 u. Similarly a reverse correction of -.16 u, will be rounded to -.20 u. Those two numbers will be added, and assuming no IOB, the "suggested bolus" will be 0.0 u. Of course, if you simply added .24 u and -.16 u, you would get .08, which you might round up to .10 or down to .05. I'm glad I understand this now (in reviewing the spreadsheet that I created to look at some of the numbers, I realized that some of my formulae were using numbers carried to three, rather than two, decimal places, which is why I couldn't see the pattern.). When we started pumping, I did look at the individual calculations, but quickly started routinely relying on the "suggested bolus". I think this rounding may be contributing to some of the wonky numbers that seem to follow going into dinner a bit below Target. I haven't looked at the numbers (not sure I need to), but we typically pre-bolus 20-25 carbs at meals, and then generally do at least one more bolus after, but sometimes, particularly at dinner, it can be more. I had been thinking that those "mini-boluses" were better, but I can see how they could add up in a way that results in DD getting a less insulin. Now that I'm aware of this, I will be more mindful of the calculations and be more ready to make changes to the suggested bolus.