Correction Factor Calculation?

I am trying to determine my correction factor, as I have to provide the info to my pump trainer. I asked my Endo’s office what it is and they said 2.5 but when I check my pump it says 3.5. I don’t use it ever, I just guess or I use it and it’s never what I think and change it.

My question is which is more conservative? 2.5 or 3.5? I’d rather err on the side of caution, as my new pump has control IQ and I think this will probably use my correction factor and for whatever reason I have NEVER been able to wrap my head around correction factors.



For CF, the number is how many points your BG is brought down by 1 unit of insulin.

I assume you are using the mmol/L numbers rather than the U.S. numbers which are mg/dl.

Is that right?

So if you are using mmol/L:

  • A CF of 2.5 means 1 unit of insulin drops your BG by 2.5 mmol/L.
  • A CF of 3.5 means 1 unit of insulin drops your BG by 3.5 mmol/L.

A CF of 2.5 is more aggressive. The reason is that it means you would take more insulin for a high BG.

Here is a quick comparison using the same BG value in mmol/L, and looking at how much you would take with those different CF’s…

If your BG is 15, and you want it to be 5…

  • With a CF of 3.5, you would take 2.85 units
  • With a CF of 2.5, you would take 4.0 units

So that shows you why a lower CF is more aggressive. Make sense?


Also known as ISF – Insulin Sensitivity Factor.


Well, mine is 40, but then, as @eric points out, I speak American.

The simplest way if you have the opportunity is to wait until you are flat-lining at an appropriately high level and do 1IU then wait to see how much your BG drops. That’s a right answer. Do it three or four times and you will see exactly how right it is. 2.5 is 45, so mine is 2.2 which is aggressive yet I still have to rage bolus. Personally I’d go with the endo number if I was using an auto-correct pump.


For a few months I would adjust the factors, set them for different types of days, and see how the BGs responded. It’s somewhat of an ongoing process although has settled down a bit.


This makes sense. Thank you!


In my experience, higher bgs can be more resistant to insulin, so correction needs to be stronger. So after calculated correction, I may do additional unit for higher bg, with carbs nearby if drops too fast.


Same. Although, I still just wing it. I want to put in the effort to test and come up with a good baseline though, then add the educated guess if I feel like it’s going to be a little strong or little weak.

1 Like

I’m a little cautious with this being my first looping experience, so I am erring on the conservative side. Ultimate goal is to be a little more brave with the correction dose, but I want to get used to the system first.

Lately, I’ve been having no end of lows and highs. I’ve been really consistent with my weight training 3x a week and more aggressive with my meal boluses, but I just can’t seem to get the numbers to flatten. The weight training piece seems to make me go high when I leave the gym. I do about 1HR weights, then 20 mins power walking. I stay fairly flat, then from the 10 min drive leaving the gym to getting home, showering, I’m suddenly about 5 points up (ie from 12 to 17). Last night, I just did cardio (long walk with my friend) and went low 3 times during, then lows in the evening and the morning. After so many glucose tabs that I felt sick from them and wasn’t going up, I had 2 pieces of sourdough toast with jam and butter, went back to bed and awoke to a lovely 18.9. Fail.


Conservative means a bigger number; telling the pump a unit of insulin will drop our BG by a greater amount. The number is backwards from this point of view. I regard my setting (40/2.2) as aggressive. Yet while it does seem to work fine for me if I’m “in range”; 100-150/5.6-8.4, it is a dead loss above 200/11; I always seem to need an additional correction later on.

This is down to the “low sensitivity at high BG” thing, but I can’t quite believe that explanation. I think it’s more a matter of maxing out the ability of our body to deal with available glucose combined with insulin-on-board degrading (so having no effect) after a few hours.

Either way the number that matters is how much our BG is decreased by 1 unit when we are approximately in range and have no glucose on board. So my metric is basically if I’m at flat at150 how much insulin do I need to get to 100. If I were to base the correction factor on higher BG I would probably over-correct when in range, resulting in a low.


I DID go bigger, my Endo said 2.5, and I chose to enter 3.5. I thought of it kinda like the assisted chin up machine at the gym–the more weight actually equates to easier. The less weight = major difficulty. ha ah it works as an analogy in my mind anyhow…

I just completed the Pump Training Video for the Tslim. While I understood everything, and got 100%, it was overwhelming. I think the alerts are going to drive me BONKERS. Anyhow, I’m going to try to just go robot mode into this no emotion.

One feature that didn’t seem like a benefit to use was the sleep function. It doesn’t give a bolus correction, but it adjusts basal. IMHO why would I use this feature? I sometimes wake up SUPER high and could benefit from a correction bolus. Other times I go low.


If I remember correctly, in sleep mode you are allowed to set a lower target BG. It will still work to lower a high BG, but by temp basals rather than by boluses. That’s not as fast, but if the pump settings are pretty nearly right, it should do the job.


The CGMs aren’t reliable; my G6 regularly drops out. If the algo delivered a bolus when I dropped out how could that possibly work? The terminology is confusing, I don’t think that’s the fault of the manufacturers, rather it is our fault for putting up with something that only applies to MDI. There is a base delivery rate, which varies over time and has a median, and there are insulin dumps. The CGMs are not adequate for automatic delivery of a dump, they only have sufficient accuracy to vary the delivery rate.

So, regardless of whether it is the O or the Medtronic or the Tandem, we have to accept responsibility for the dump and the pump handles the base in (continuous) response.

1 Like

The US FDA came to a different conclusion when they determined that the t:slim Control-IQ algorithm is safe and effective.

1 Like

I deal with this by using cgm alarm at 140-150. If alarms, then i do manual bolus.

But agree, many users complain about no option to update user setting for this.

One option is to tweak basal settings to be slightly higher, and have it reduce/suspend rather than chasing highs with minimal basal increases.

1 Like

So it will do a dump? You said it didn’t in sleep mode, so when does it do one?

Below is detail from Tandem.

During day, Tandem relies on user set high/low alerts to help stay in lower range via user corrections. With sleep mode, it uses basal adjustments successfully most of the time during fasting nighttime. But many times the pump basal settings don’t really match every situation during nighttime, but in general they are more predictable than daytime.
I use sleep mode 24x7, and rarely get a CIQ bolus “dump”.

The bolus correction is based on trend prediction.
Prior to bolus, there would already be extra basal working to slow bg rise or start decrease. Bolus is less than 100% to accommodate that additional “IOB” from increased basal.


Perhaps just 3/4 pice of sour dough without the jam? Moderation in all things. We’re all different so my experience may not apply, but I’ve learned (and tried to apply…not always successful) the lesson that when I go lower than desired (<70 and dropping) to moderate the carb intake that I’d “like” to apply, a little at first, wait 5-10, a little more, wait 5 again for an upward inflection…understanding all the while my CGM is 10-15 behind…and applying how I “feel” along with the numbers. It may be a crap shoot…but it’s my crap shoot.


Yes, the O approach is to incorporate IOB into all the calculations, so a bolus (== single delivery of a large quantity of insulin, dump) is never going to happen unless the basal (== the rate at which which insulin is delivered continuously, base rate) is interrupted by food (which messes up the CGM measurements).

Tandem are using a BG range which, by my standards is very small:

160 mg/dL was calculated to be a safe and effective glucose level to begin to increase basal insulin delivery to help prevent going above 180 mg/dL. If glucose continues to rise and is still predicted to be above 180 mg/dL in 30 minutes, Control-IQ technology can add an automatic correction bolus to help minimize time spent with high glucose.

It’s not clear what the O uses; I think it might start outside +/-20mg/dL but I could not find a definitive statement when I skimmed the user manual.

In any case the whole C-iQ approach strikes me as overcomplex; instead of just increasing the base rate C-iQ also delivers single doses every 60 minutes. I sometimes do that but that’s because I’m not a CGM; I’m not checking every 5 minutes and because I don’t correct until 180mg/dL. Even then I sometimes do this by adjusting my base rate, particularly if I’m sleeping.

If I were a CGM I would produce an internal model of how increases to base rate effect changes to rate-of-change of BG then use that to derive an appropriate change to base rate. In the absence of that I might use the C-iQ algorithm to deduce a single dose but rather than arbitrarily reducing it by 40% I would remove the current base rate adjustment (i.e. the amount above actual basal requirements) and replace it by the single dose spread over 30 minutes; a base rate change. I do sometimes do that at present.

Is my understanding correct that although Sleep Mode will not give a correction bolus, it will raise and decrease basal as needed? Just trying to wrap my head around why people stay on Sleep Mode.

1 Like