The trouble with pump reverse corrections

[saxmaniac]

I've been doing a little bit of analysis of our pump's reverse correction algorithm. I could give a lot of details, but I know it will bore most people here. I'll post that later. Right now I'd like to know if you have the same issue!

Suffice to say, I think it's inaccurate. Our pump does the correction in terms of insulin, not in terms of carbs.

So, when it wants to raise you 50 points (say from 70 to 120) it will shave of 50 points worth of insulin. That amount comes from your ISF and I:C. And so the amount changes as the ISF and I:C change over the day.

But the problem is there's no insulin for those carbs!

Let's say the pump will do a correction, and Alex is planning on eating. It's going to give me a different carb amount based on what time of day it is, because it's using my ISF and I:C to calculate a dose first. This doesn't match with my experience, that a certain amount of carbs raises the BG the same, all day and night.

When we are on MDI we always did it in terms of carbs. If I needed to get 50 points, and there's no IOB, then it would be about 8g. All day. No matter what the ISF or I:C is. So we'd add 8g to the meal (and not bolus it), or subtract out 8g from the dose. I never read this anywhere, just arrived at it from our own experience.

The pump inaccuracy is more pronounced depending on how far you are away from your target. Whether you err on the side of too much insulin or too little depends on your I:C and ISF, compared to your carb sensitivity.

Anyone have this issue? Do other pumps calculate reverse corrections the way I used to do it?

[twodoor2]

HAHAHAHA - the infamous negative correction issue that I warn people about. I think it causes more glucose variabilty than it's worth. I'm not a fan of negative corrections and pumps do this when under target range. If you do this, you also have to make sure that your ISF is exact, and how many endos really pay as much attention to ISF (let alone variable ISF's) as they do I:C ratios?

http://forums.childrenwithdiabetes.com/showthread.php?t=15800

I think your way is much safer to do.

[BrendaK]

When we quit doing negative corrections on the Cozmo pump, our lives were GREATLY improved. I agree completely with your theory.

I just filled out paperwork for Carson to go to D camp (this is the second batch of paperwork they want!) and it wanted to know what our negative correction factor was. I had to explain on the form we don't use a negative correction. If he is low at mealtime, then treat with 2 glucotabs, THEN bolus for all the food. I hope they will accept that, because I didn't have a formula to give them.

[thebestnest5]

We just treat with food and don't use the negative correction for the same reasons.

I couldn't make sure her ISF was exact every minute of every day due to the variables that I have no or almost no control over.

The only time, that we use negative corrections if when Liv is with friends and I would rather she run on the higher side of normal for her own safety. So, it has been helpful to us in very limited circumstances. Such as now that Liv is out and about with friends and doing things more independently. This week she went to Nim's Island & shopping with a group of friends and the mom who drove asked if it was OK, if the girls attended the movie and she picked them up. Liv's sister was also with and I had them text me. All went well and I told Liv "No,(high) corrections with insulin...just drink water if you want and we'll correct when you get home". Also, the MM sensor was so helpful since it was reading within 3%:D of her BG meter.

[Mrs. Russman]

We've had the same problem...
I finally told Ben to not enter lower blood sugars, and just subtract 10 carbs from the total carbs being eaten.
The other morning he was at 64, and the pump recommended 0 units for 33 carbs! of course within an hour he was well over 200.

[Twinklet]

Same issue here! We NEVER use the negative correction feature on the Cozmore. It is WAY wrong. In fact, there are times when I really think they subtract both insulin AND carbs from the formula when there's existing IOB. Emily would be sky-high later on if we used it. It makes me crazy.

We just shave off however many carbs we need to from the meal, unless she's low and we need to treat with tabs.

[Treysmom]

Hand raised..........:o

Are you talking about below 70 bg's here. I'm getting part of this. I need alittle more help.

I do treat a below 70 # 1st by shaving carbs or giving glucose tabs. Then feed.

[saxmaniac]

Treysmom - no, we're talking below target but still in range. Our pump is set to allow bolus calcs above 70. Above that and below the target (120 day) it will do a negative correction.

Thanks for the responses! I am going to stop using it for a bit and see if it helps. Anyone else have good results and leave it on?

[albasmom]

I hadn't thought of this. We'll try and stop using it too and see if it helps.
Thank you for thinking of this :)

[Treysmom]

I get it now. My brain is running a little slow today. Went out to eat at last night to much heavy italian food. :o

[twodoor2]

Same issue here! We NEVER use the negative correction feature on the Cozmore. It is WAY wrong. In fact, there are times when I really think they subtract both insulin AND carbs from the formula when there's existing IOB. Emily would be sky-high later on if we used it. It makes me crazy.

We just shave off however many carbs we need to from the meal, unless she's low and we need to treat with tabs.

I think this problem is even more exaggerated with the Cozmo and Omnipod due to the linear IOB.

[hawkeyegirl]

What I don't get is why the pump companies haven't figured this out. We, as dumb old layperson parents know that the current system doesn't work. Why don't they fix this? (Same thing with the Minimed dropping the hundreths place when it does its calculations. This is not tough. These are $6,000.00 machines.)

[twodoor2]

What I don't get is why the pump companies haven't figured this out. We, as dumb old layperson parents know that the current system doesn't work. Why don't they fix this? (Same thing with the Minimed dropping the hundreths place when it does its calculations. This is not tough. These are $6,000.00 machines.)

I suspect that the negative corrections and as you mention, the truncation of the hundredth place in the MM522/722 series, work for adults. Adults typically have very low target ranges as well, so the chance that they'll be subtracting insulin is not as often as perhaps it is for a child. Small children usually have higher target range minimums. Adults tend to be also much less sensitive to insulin, so maybe subtracting a unit here or there doesn't really affect them so adversely.

My wish is that the pump companies would offer more suitable algorithms for children, in particular, very small children that need specialized pediatric models. I like that you can "turn off" the negative correction feature on some pumps. The only way to do this on the MM522/722 series is to lower the target minimum to a point where negative corrections do not come into the picture that often.

Elizabeth's target range is 80 - 140, so it's rare for her to get a negative correction, she would have to be < 80. When she does, she's in the 70's. It works well when she's this low, because the negative correction amount is very small - (70-80)/ISF. It causes more issues when she's higher because the numerator would be larger and the subsequent negative correction would be larger.

[Jacob'sDad]

I've been doing a little bit of analysis of our pump's reverse correction algorithm. I could give a lot of details, but I know it will bore most people here. I'll post that later. Right now I'd like to know if you have the same issue!

Suffice to say, I think it's inaccurate. Our pump does the correction in terms of insulin, not in terms of carbs.

So, when it wants to raise you 50 points (say from 70 to 120) it will shave of 50 points worth of insulin. That amount comes from your ISF and I:C. And so the amount changes as the ISF and I:C change over the day.

But the problem is there's no insulin for those carbs!

Let's say the pump will do a correction, and Alex is planning on eating. It's going to give me a different carb amount based on what time of day it is, because it's using my ISF and I:C to calculate a dose first. This doesn't match with my experience, that a certain amount of carbs raises the BG the same, all day and night.

When we are on MDI we always did it in terms of carbs. If I needed to get 50 points, and there's no IOB, then it would be about 8g. All day. No matter what the ISF or I:C is. So we'd add 8g to the meal (and not bolus it), or subtract out 8g from the dose. I never read this anywhere, just arrived at it from our own experience.

The pump inaccuracy is more pronounced depending on how far you are away from your target. Whether you err on the side of too much insulin or too little depends on your I:C and ISF, compared to your carb sensitivity.

Anyone have this issue? Do other pumps calculate reverse corrections the way I used to do it?

I don't know. I'll have to check this out more closely and get back to you. I for one do not think that 1g of carbs raises BG the same day or night. Why would it? Jacob is generally more active in the middle of the day and that activity helps keep his BG down, so I believe that 1g of carbs is going to raise his BG less then than it would at 3am.

[Gwyn]

I have noticed this too.

When we treat a low, we usually figure 1 carb raises BS 8 points. When we treat a high, we use BS-target/5--and then cover the result as carbs (not eaten) at the appropriate ratio for the time of day.

Note to Marsha: This is where I got mixed up in my example that one time. I was thinking at the time that I was just being more cautious when treating lows, but what saxmaniac is saying here is a much better argument.

[Gwyn]

I for one do not think that 1g of carbs raises BG the same day or night. Why would it?

I think that 1 gram carbs raises BG the same every time--*all else being equal*. Of course we know that there are always other forces at work (exercise level, IOB etc.), so it makes it difficult to isolate that particular variable. Practically speaking, I think your assumption works, but it is likely not true in a strict sense.

IMO that's the most aggravating thing about number crunching and diabetes--there are so many interactive variables it makes my head explode.

[Jacob'sDad]

I think that 1 gram carbs raises BG the same every time--*all else being equal*. Of course we know that there are always other forces at work (exercise level, IOB etc.), so it makes it difficult to isolate that particular variable. Practically speaking, I think your assumption works, but it is likely not true in a strict sense.

IMO that's the most aggravating thing about number crunching and diabetes--there are so many interactive variables it makes my head explode.


I just think that the same NORMAL level of activity in the middle of the day that makes his lunch I:C higher also makes his BCR lower. I can't consider one in a resting state unless I consider BOTH in a resting state.

[hold48398]

I am so totally with you on this one!!!!!!!!!!!!!! Never use neg correction here but always correct in terms of carb deductions.

[saxmaniac]

I think this problem is even more exaggerated with the Cozmo and Omnipod due to the linear IOB.

Maybe, but I don't see it. If he's 80 before eating, there's usually no IOB, as IOB only comes from boluses w/o carbs (corrections, normally). It's 0 a lot, or so close to 0 as not really to matter.

[twodoor2]

Maybe, but I don't see it. If he's 80 before eating, there's usually no IOB, as IOB only comes from boluses w/o carbs (corrections, normally). It's 0 a lot, or so close to 0 as not really to matter.

I don't know how the omnipod works, but the MM takes IOB into account regardless of a straight BG correction or not. In fact, if you are giving a correction, it will always take IOB into account, unless there is none left. As for the fact that he's 80 and there's no IOB left, as we all know, you can still have substantial IOB left and still be in the low target range or hypoglycemic. That's why it's important to gage BG relative to IOB.:)

[saxmaniac]

I don't know. I'll have to check this out more closely and get back to you. I for one do not think that 1g of carbs raises BG the same day or night. Why would it? Jacob is generally more active in the middle of the day and that activity helps keep his BG down, so I believe that 1g of carbs is going to raise his BG less then than it would at 3am.

IMO the insulin is negating the carbs there, and it only makes sense to me to consider how much carbs raise the BG when everything is steady. (Yeah, which never happens! :D )

You give 1g of carbs, and it goes up less because other factors are bringing it down. Many times I give a carb correction to stop a steep downslope, it doesn't bring him up, but it stops him from going down!

So I think of it as, say, 50 points that got "eaten up" by the insulin, but it still went in there.

Otherwise, if we considered how the insulin works against it, there would be times when I give 5g of carbs, and it would go down 20 points! If that's the case then I'd stop buying insulin!

Having this number is useful. I know if he spikes a lot more than the carb amount says it should (6.6/g) then there's likely something else at play: a big fat spike, basals are off, bad site, etc.

[BrendaK]

Maybe, but I don't see it. If he's 80 before eating, there's usually no IOB, as IOB only comes from boluses w/o carbs (corrections, normally). It's 0 a lot, or so close to 0 as not really to matter.


Cozmo takes into account IOB from corrections AND meal boluses. I wish it would only take into account IOB from corrections and not the meal boluses, because food is usually still digesting.

So if Carson is 80 before eating, the pump will subtract insulin on board from the last meal (DIA is 4 hours) and it will subtract insulin because he is below target (120). So in a sense, it "double dips" when it comes to negative corrections. Doesn't work for us.....