Insulin on board: pumps and their bolus wizard

Treatment Insulin
1

Version 2.0: I did some rewording. Thanks for the feedback!

Insulin on board (IOB) is a feature of most modern insulin pumps. For our Glucosurfer project we integrated this useful estimation too. But our take on the IOB is a little bit different:

OmniPod IOB:

In a perfect world we always count our carb correctly to the gram. In combination with a perfect I:C the injected insulin and digested carbs will neutralize each other. Just an additional correction for a previous blood glucose that was elevated needs our attention. Here the OmniPod just calculates the IOB from the previous corrections. With this approach it will prevent that we will add another correction on top when the next blood glucose is still elevated.

Perspective: in theory this is correct but practically we all have different experiences with huge loads of insulin on board. Just additional physical activity can amplify the effect of insulin for example. So the total has an important information in itself that is missing here.

Medtronic IOB (or Cozmo IOB):

They count all the previous injections for the calculation of the IOB.

Perspective: here the IOB acts as a good reminder of all the insulin that is still active.

Flaws in the bolus wizard:

Lets imagine you have 150 mg/dl now and you want to eat 30g of carbs. The bolus wizard will calculate a correction of 0.5 plus 30/15 = 2 => 2.5 units in total.

This has been clarified now: Pumps like the Medtronic will calculate the total IOB. So the IOB might warn that 3 units are still active. But then and this is interesting it will subtract only the correction part of the IOB from the calculated dosage like this: 2.5 - 0.5 = 2 units (if 0.5 units where left from a previous correction). What happens if there was previously no correction? Will it then suggest a total of 0.5 plus 30/15 => 2.5 units?

On the OmniPod the IOB of 3 would mean that correction insulin is still active. So the 150 mg/dl would not need an additional correction. The result would be just 30/15 = 2 units for the carbs to digest. The problem I see is that this number is always subracted from the total the bolus wizard has calculated not from the correction estimate alone.

Compromise is bliss:

1. show the total IOB.

2. apply the IOB to the calculated correction ONLY and reduce it to 0 if neccessary. This is actually as I have just learned like the Medtronic pumps are working.

Again the example: 150mg/dl now, 30g to eat, IOB of 3 from the previous meal. This will lead to 30/15 = 2 units for the carbs. The correction of 0.5 for the 150mg/dl will be reduced to 0 according to the IOB of 3. Here the nature of the IOB (just for carbs, just for correction, correction plus carbs) will not matter anymore.

This is how we calculate and use the IOB for our Glucosurfer App. In my opinion this combination of both worlds offers the best approach. It acts as a warning and prevents the stacking of corrections at the same time.

Are you all aware of these potential problems? Have I been to general in my evaluation? Can you make adjustments to your pump algorithms to make them work like I suggest here?

2

Your info is kinda misleading, the Medtronic pump tracks total IOB but only adds or subtracts correction insulin based on target settings and how much correction insulin is still on board from total correction insulin, all carb insulin is always covered by the carbs entered into the pump. The total IOB is shown based on a curve that is similar to insulin action times of humalog and novolog.

I also think your explanation of how the OmniPod tracks IOB is incorrect unless they have made a change to their bolus calculator it only tracks correction insulin and does not display active carb insulin like a MiniMed pump or the old Cozmo pumps. ...JMO

3

Thanks John, just for clarification with the Medtronic way:

Then the IOB is calculated as total like the Cozmo did. Let us assume without IOB the correction would be 0.5 units. What happens if there is insulin from carbs on board and previously no correction has been applied? Will the Medtronic still suggest to inject 0.5 units although there is insulin on board?

As a side note: actually the curve for IOB calculation is the same for all insulins. It is just the duration of the insulin activity you provide in your settings that will adjust the curve to the duration of your insulin.

I will change the wording of the discussion...

4

I am not sure I completely followed your thoughts, Holger. In particular, I am struggling with two things. The first is this from your post:

Perspective: ... Here I am really missing to be warned about that with a high IOB number.

A translation problem?

The second area is how IOB is calculated. Straight line or using some sort of algorithm?

On my Accu-Chek Combo, the Bolus Advisor calculates a straight line equation based on how long I say that Humalog is active in my body. So if I say Humalog is active for 4 hours and I have taken a bolus of 3 units, the wizard will say I am using 0,75u/hr. This results in the wizard recommending the wrong amount of insulin if I need to bolus for additional carbs within the 4 hour window. I typically end up doing a little calculation in my head that starts with what the wizard thinks about IOB. So in my opinion, the wizard/advisor is not a very useful tool. (As an aside, I changed the active period to 3 hours to force the wizard into less IOB from 2,5 - 4 hours after bolus.)

5

I reworded my discussion. Hopefully it is a little bit clearer now.

In my opinion they should all use the same curve (adjusted with the duration). The difference to the straight line is significant as your example shows. But the intended focus is more:

a) what the IOB really is to some manufacturers
b) how this IOB is influencing the calculations of the bolus wizard

6

Now that I re-read your compromise example, I think this would create a low in my experience. This is because with a 1:15 I:C ratio, your 30 carbs would need 2 units with 0 IOB. Therefore, I don't see where you have taken IOB into account.

7

Mike, that is the interesting part: with an IOB of 3 we just prevent that the 150 mg/dl will be corrected to the target. For the additional carbs we inject normally.

I think this is the core of the discussion. Why should the second meal be treated with less insulin? From the previous meal we have carbs and insulin on board. They should neutralize each other. To take the correction out of the equation is rational too. But why fearing a low? It is your experience obviously and it would be interesting how this is seen by others.

8

There is 3 diffrent ways that a Medtronic pump will determine a bolus.

1 If your current BG is greater than your High BG Target, the Bolus Wizard feature subtracts active insulin
from the BG correction estimate, then adds this to the food estimate to get the total bolus estimate.
However, if the result of subtracting active insulin from BG correction estimate is a negative number
(less than zero), the total bolus estimate is based only on the food estimate.
Food estimate:
Carb grams ÷ Carb ratio = Units of insulin
Correction estimate:
(Current BG - High BG Target) ÷ Insulin sensitivity - Active insulin = Units of insulin
Total bolus estimate:
Food estimate + Correction estimate = Units of insulin

2 If your current BG is less than your Low BG Target, the Bolus Wizard feature adds the BG correction
estimate to the food estimate to get the total bolus estimate.
Food estimate:
Carb grams ÷ Carb ratio = Units of insulin
Correction estimate:
(Current BG - Low BG Target) ÷ Insulin sensitivity = Units of insulin
Total bolus estimate:
Food estimate + Correction estimate = Units of insulin

3 If your current BG is between or equal to your High or Low BG Target, the total bolus estimate is based
only on the food estimate.
Food estimate:
Carb grams ÷ Carb ratio = Units of insulin
Total bolus estimate = Food estimate

9

Thanks for your detailed explanation: your explanation is actually how the Glucosurfer is working. With the separation of food and correction estimate. With applying the active insulin on just the correction estimate. With the consideration of neutral carbs you can calculate for lower BG values.

Am I correct that "Active insulin" is always the total IOB?

10

I have better scenarios to show the differences:

Szenario 1

One hour ago a BG of 250 mg/dl has been corrected. The total IOB for Metronic and the correction IOB on the OmniPod will both show 2.5 units on board.

Now the blood glucose is 150mg/dl and we want to eat 30g of carbs.

Medtronic:
carb estimate: 30/15 = 2
correction estimate: 1 - 2.5 = 0
total: 2

OmniPod:
carb estimate 30/15 = 2
correction estimate = 1
total: 3 - 2.5 = 0.5

Opinion:
The Omnipod acts very cautious here - maybe a bit too cautious for my taste. The carbs of the second meal might get less insulin than needed.

Szenario 2

30 min ago a BG of 180 mg/dl has been corrected (2 units) and 60g of carbs have been eaten (4 units). The total IOB for Metronic shows 5.8 units left of 6. The correction IOB on the OmniPod shows 1.8 units left of 2.

Now the blood glucose is 170 mg/dl and we want to eat 30g of carbs (not very wise I know).

Medtronic:
carb estimate: 30/15 = 2
correction estimate: 2 - 5.8 = 0
total: 2

OmniPod:
carb estimate: 30/15 = 2
correction estimate: 2
total: 4 - 1.8 = 2.2

Result:
Here the results are comparable.

It really depends on the scenario to find minor to major side effects of the different strategies. In my opinion the different strategies for IOB calculation and its use in the wizard should be adjustable by the user. If we have picked our pump but dislike the IOB strategy we are locked-in (unhappy customers). This will also prevent users of other pump models to switch to another vendor if the strategy they are accustomed to is not supported (vendor lock-in). So this adjustability will help pump vendors too.

11

From another discussion I have learned that OmniPod just recently changed the IOB calculation from just correction IOB to total IOB. The PDM has been updated with the new OmniPod models. The remaining question is if the use of IOB in the bolus wizard is now comparable to Medtronic's wizard (John described that in detail in this discussion)?

12

Interesting topic, Holger. Thank you for bringing this up.

I have had pumps from Medtronic for more than 10 years and always been satisfied with their software "Bolus Wizard" and how it works also regarding to IOB. I got my first pump from Animas this june, but I still can´t figure out is how their software ezCarb and ezBG works regarding the IOB-issue. Do you have any info on this? And is it possible to throw Animas into the discussion and compare their method of IOB-calculation to the others?

13

I think the Animas Ping does calculate a total IOB. But the use of the IOB in the bolus wizard is different from Medtronic's approach. Because the ping will subtract the total IOB from the carb portion of the bolus IF YOU ARE BELOW TARGET RANGE.

To some degree I can understand this logic but I do not like it. Being lower than the target is interpreted by the Ping as if there are no more carbs to digest OR all carbs have been digested and now there is too much insulin left. But with slowly digested carbs or slower digestion these lower numbers in the first two hours might be normal. In this situation the ping might subtract too many units from the carb estimate of the following meal (in my opinion). To my knowledge this can be only be compensated by reducing the level for low numbers in the ping settings. But I am not sure about the negative side effects.

I have no ping or its manual at hand. So my conclusion might be based on wrong assumptions. Please correct me if I am wrong.

14

Thanks, Holger. I have no more info than you on this. Since coming back to insulin again it´s been enough just keeping my head above water. Seems my brain has gone into slumber mode. Hence the question. If someone could explain it for my like you would to a child, I´d be endlessly happy.

That said, - I always do my own calculations in addition to what my pump tells me, and usually ends up overriding the pumps calculations. I seldom did that when using a Medtronic pump, and I just can´t figure out how this Animas pump actually "thinks" and how it doffers from Medtronic.

15

Is it just the IOB? Or do you have to correct normal calculations too?

16

Holger,

The new OmniPod and PDM have changed their IOB calculations...they now include meal IOB.
I just switched and I'll have to see how it works out. The old Omnipods and PDMS are being discontinued.

The new manual is located here:
https://www.myomnipod.com/pdf/14421-AW%20UST400%20User%20Guide%20Rev%20C.pdf

The IOB calculations are on page 143 - 146 in the appendix.

Meal IOB is included in total IOB and is used to reduce correction IOB and only reduces the correction down to 0 now. Meal IOB is never used to reduce another meal bolus.

17

Thanks for link to the manual. In another discussion I have already learned about this important update for the PDM.

Does the OmniPod know about neutral carbs? These are carbs you could eat to raise your current BG to the target. This has nothing to do with the IOB. It just acknowledges the fact that for example with 70 mg/dl you could eat 10g of carbs - and this without additional insulin. Below target we subtract these neutral carbs for the carb estimate of the bolus wizard of our Glucosurfer App. There is always room for refinements...

18

Active is the total IOB, but this is sort of a "fudged" number as you are estimating 3 or 4 hours or whatever your "duration" is but there may still be "stray" insulin floating around. The pump thinks it knows everything but it sort of only knows what you tell it.

I keep my target set pretty close, 85 for a low and 90 for a high. I think I might have perceived this in "Think Like a Pancreas" whie flipping through it and it's in keeping with my suspicion that a smaller target will improve my results. That way even the lower 100s will get a "whiff" of correction. These small doses don't seem to do a lot in terms of causing hypos or anything like that but I suspect they might keep some 120s from turning into 130s or 40s to help cook my numbers in the right direction.

19

There is no explicit recognition of this now on the Omnipod as far as I can tell.

Some of the books on insulin management suggest that those that can't do this calculation in their heads can use the pump bolus calculator indirectly by trying trial carb inputs until the bolus recommended is zero.

The Omnipod does do reverse correction if you turn it on....i.e. it will subtract insulin from the recommended bolus (both meal and correction) if you are below target.

20

Hurray to the developers for making the reverse correction optional!